In 1970, after a 4-year effort, the Commission on Classification and Terminology of the International League Against Epilepsy (ILAE) presented a scheme for classifying epileptic seizures, in which the term “partial seizures with complex symptomatology” was introduced.⁵⁴ This term was used to denote a type of seizure that was clearly described almost 100 years earlier by Hughlings Jackson,⁷⁴ who noted a relationship between lesions in the mesial temporal lobe and seizures characterized by olfactory hallucinations and a “dreamy state,” which he referred to as “uncinate fits.” Since then, myriad signs and symptoms, both psychic and motor, have been identified with this seizure type. In 1937, Gibbs et al.⁵⁶ proposed the term “psychomotor” to define these phenomena. Subsequently, the Montreal school, through their pioneering invasive recording and surgical treatment efforts, repeatedly demonstrated the clear association between psychomotor seizures and temporal lobe pathology, changing the terminology to “temporal lobe seizures,” as part of a condition known as temporal lobe epilepsy (TLE).^75,104 Because the limbic system primarily involves mesial temporal lobe structures, the specific condition is now known as mesial temporal lobe epilepsy (MTLE).¹⁴⁰

The advent of long-term video electroencephalographic (video-EEG) recording allowed routine electroclinical correlation and documentation of habitual seizures, permitting detailed analysis of ictal semiology. It became recognized that not all seizures of temporal lobe origin have psychomotor features; some seizures have psychic symptoms without motor signs, and others have motor signs without psychic symptoms; seizures originating outside the temporal lobe can propagate into limbic areas, producing ictal events that are indistinguishable from temporal lobe onset seizures; and seizures originating outside the temporal lobe could have complex symptomatology without involving temporal limbic structures. Whereas the 1970 International Classification essentially used the term complex partial seizure (CPS) as synonymous with “psychomotor seizure” or “temporal lobe seizure,”⁵⁴ the revised 1981 Classification²¹ chose to define terms purely on the basis of phenomenology and to avoid any implication of anatomic substrate. Consequently, CPSs were defined as partial seizures associated with impairment of consciousness, as opposed to simple partial seizures (SPSs), which are partial seizures with preserved consciousness.

Now that more is known about the pathophysiologic mechanisms and anatomic substrates of various epileptic seizure types, it is recognized that classical temporal lobe seizures, beginning in mesial temporal limbic structures, can be simple or complex partial, whereas ictal impairment of consciousness can occur during partial seizures that arise from the neocortex and never involve mesial temporal limbic structures.¹⁴⁷ Because modern electrophysiologic and neuroimaging technology now permit accurate anatomic localization of ictal events in many patients under evaluation for surgical treatment, it has become common to distinguish between limbic partial seizures and neocortical partial seizures based on their characteristic ictal semiology. Consequently, the ILAE has now recommended that SPS and CPS not be used in a future classification, but have accepted the term “limbic seizures” to describe ictal clinical manifestations of epileptic activity in mesial temporal limbic areas and their efferent projections.^37,39

Although the current 1981 International Classification of Epileptic Seizures appeared to satisfy many of the needs of the epileptology community, confusion and lack of acceptance persisted among some practicing neurologists because highly complicated and sometimes bizarre seizures in which consciousness was not impaired were, by definition, simple partial seizures. Conversely, patients with partial seizures who have only a motor arrest with impaired consciousness, by definition, have CPS, which can be mislabeled as petit mal. Finally, although the term “complex partial seizures” has been universally accepted, it continues to be abused by reversal to the meaningless term “partial complex seizure.”^{3,44,71,72,103}

Defining consciousness is difficult.^11,57,58 For example, seizures can present with a relatively pure reversible Wernicke aphasia. In this situation, the patient would appear alert and awake. Speech would be jargon, and commands would not be followed. On the other hand, patients can present with seizures consisting of pure word muteness or aphemia,¹⁴⁶ during which the patient would be able to follow commands and perform all other language-related functions, but simply could not speak. Are these examples of CPS or SPS? The latter should be considered simple partial seizures, but the former could be difficult to classify when impairment of receptive language function complicates testing consciousness. Amnesia for the ictal event is another criterion that can be used to define CPS.³³ Amnesia, however, whether complete or partial, can be as difficult to document as altered consciousness. Careful testing by trained observers during and after seizures is the only method that can define these issues,¹⁴¹ but in practice, this has not been standardized, nor is it universally performed in all centers where seizures are recorded. Although recognition of whether consciousness is impaired is of clinical importance in individual patients, the ILAE has recommended that this difficult-to-assess clinical manifestation not be used as a criterion for classifying seizure types in the future.^37,39 For purposes of describing the ictal manifestation of cognitive dysfunction, the ILAE has now recommended the term “dyscognitive.”¹¹

The term “limbic seizure” is used in this chapter to describe ictal events in which the clinical manifestations result from epileptic activity within the mesial temporal limbic network and structures to which it projects. Limbic seizures can be initiated by epileptogenic regions within this limbic network, or from epileptogenic regions in any neocortical area that projects to it and gives rise to the same clinical manifestations. Epileptic activity will preferentially propagate to mesial temporal limbic structures from the temporal neocortex, orbital frontal cortex, mesial frontal regions such as the anterior cingulate, and
occipital cortex ventral to the calcarine fissure. Certain sensory, psychic, and autonomic manifestations of limbic seizures do not usually involve impairment of consciousness, and so would currently be classified as SPS.

Most prevalence studies estimate that approximately 60% of the adult population with epilepsy and 40% to 50% of children with epilepsy have focal seizures.^{18,24,25,50,55,68,79,80,89,103} The earlier Rochester study by Hauser and Kurland reported that CPS were the most prevalent of any single type of seizure, constituting 42% of focal seizures and 26% of all seizures. Another earlier study reported that 40% of patients with epilepsy had CPS as the predominant seizure type.⁵⁵ Epidemiologic studies from developing countries usually report much lower percentages of patients with CPS.^{2,28,47,64,77,118,126} The reasons for this discrepancy are not entirely clear but probably reflect such factors as diminished diagnostic accuracy and different etiologies. However, the inherent difficulties of seizure classification exist at any level of societal development.^92,102,132

CPS are not all limbic seizures and not all limbic seizures are CPS; therefore, attempts to further subclassify focal seizures based on limbic system involvement are at best an estimate. By necessity, such approximations come from epilepsy surgery centers, where ictal substrates are most thoroughly and precisely evaluated and where surgical results give credence to the localization of seizure origin. This introduces an undeniable referral bias that cannot be avoided. Of those patients referred for surgical consideration, patients with limbic seizures of temporal lobe origin are by far the most common, constituting approximately 70% of the referrals.^138,151 Of the other 30%, approximately two thirds are patients with different varieties of frontal lobe seizures, with the remaining 10% divided among seizures originating in other lobes of the brain.^{107,139,143,150} Some of these patients also have seizures generated by the limbic system, even though they do not originate in limbic structures.

Many pathologic conditions, idiopathic as well as symptomatic, primarily involving limbic structures or occurring in distant neocortex, can be associated with limbic seizures as the predominant ictal features. These conditions are discussed in detail in Chapters 246 and 247. The most common pathologic substrate associated with limbic seizures is hippocampal sclerosis (HS) (Chapter 13), which can now be detected in most patients using high-resolution magnetic resonance imaging (MRI).^73,83 Another indication of the prevalence of limbic seizures, therefore, can be derived from two large longitudinal outpatient studies, one in Paris that included tertiary referrals,¹¹⁶ and the other at a primary clinic in Glasgow.¹²⁸ In the Paris study, half the patients had a diagnosis of TLE and, in both studies, approximately a quarter of the patients had hippocampal atrophy documented by MRI. In both studies, the seizures in patients with hippocampal atrophy were the most refractory. In the Paris study, only 11% were seizure-free in the previous year, and with dual pathology (hippocampal atrophy plus another MRI abnormality), only 3% were seizure free in the preceding year.¹¹⁷ Although there is no way to accurately assess the true prevalence of limbic seizures and the burden of disability due to epilepsy that limbic seizures represent, it is reasonable to assume that they are the most common, and are certainly among the most refractory epileptic seizure types.

Because patients with TLE are the most common candidates for surgical treatment,^45,46 and often require invasive V-EEG monitoring, many more limbic seizures have been explored with stereotactically implanted depth electrodes in TLE than any other seizure type. It is clear from these recordings that ictal activity that remains confined to hippocampus and parahippocampal structures can have no clinical manifestations at all,¹²⁵ and that the classical signs and symptoms of limbic ictal events reflect propagation, both ipsilaterally and contralaterally, to frontal and temporal neocortex, insula, hypothalamus, basal ganglia, and other subcortical structures. Although limbic and neocortical propagation patterns have been explored with depth electrodes,^81,85,87,124 and thalamic and basal ganglia involvement has been inferred from ictal behavior as well as structural and functional neuroimaging,^8,13,70,101 the contribution of other subcortical structures to the typical signs and symptoms of limbic seizures in the human cannot be directly studied.

Both animal and human investigations suggest that the pathophysiologic mechanisms responsible for seizure initiation in the hippocampus, and perhaps parahippocampal structures, may be different from those of seizures originating in neocortex;^16,114,133 this is discussed in Chapter 41.

The limbic anatomy and physiology responsible for the elaboration of clinical signs and symptoms of limbic seizures are described in Chapter 30. As previously mentioned, ictal discharges restricted to hippocampus alone are often devoid of noticeable clinical manifestations.¹²⁵ Ictal symptoms that occur in clear consciousness—the typical limbic auras—result from usually ipsilateral projections. Autonomic symptoms and emotions such as fear most likely reflect propagation to hypothalamus and insula, whereas complex multimodal sensory psychic experiences result from neocortical projection, particularly the temporal parietal occipital junction.^59,129 Limbic structures responsible for memory, as well as taste and smell appreciation, give rise to dysmnesias and olfactory and gustatory auras. More complex motor symptomatology, characteristic of limbic automatisms, however, is almost always associated with impaired consciousness and usually involves contralateral propagation. Bilateral hippocampal involvement accounts for ictal amnesia, although unilateral ictal discharge can produce amnesia when the contralateral hippocampus is severely dysfunctional. Oroalimentary automatisms derive from the temporal and frontal operculum but the specific origin of more complicated gestural automatisms has not been clearly defined.

Studies in animal models of MTLE with HS reveal strong inhibitory mechanisms within the hippocampus that suppress contralateral propagation.⁹¹ A similar resistance to contralateral propagation exists in patients, as evidenced by the common observation that depth-recorded limbic ictal discharges remain unilateral for long periods of time and may never involve the contralateral side.^85,87 The existence of intrinsic mechanisms that permit ictal discharges within the limbic system to remain unilateral for extended periods of time without propagation likely accounts for the clinical observation that auras in patients with limbic seizures of mesial temporal origin commonly occur in isolation; these patients typically experience auras that do not progress to behavioral seizures. In contrast, neocortical auras rarely occur in isolation because of the propensity for neocortical discharges to rapidly project contralaterally across the corpus callosum. When contralateral propagation occurs, it is usually indirect via either temporal or frontal ipsilateral neocortex.^{81,85,86,87,124} Because ictal discharges can be seen in contralateral mesial temporal limbic structures before involvement of contralateral neocortex, the route of contralateral propagation may not be callosal. Furthermore, it is not possible to explain how bilateral automatisms can occasionally occur in clear consciousness without amnesia.

The neuronal networks responsible for the manifestations of limbic seizures involve neocortical and subcortical areas.
Extensive extrahippocampal cortical and subcortical gray matter atrophy may be apparent on MRI,^8,13 and positron emission tomography (PET) often reveals unilateral thalamic and basal ganglia hypometabolism ipsilateral to the side of ictal onset in patients with limbic seizures.⁷⁰ Unilateral basal ganglia hyperperfusion on single-photon emission computed tomography (SPECT) is associated with contralateral ictal dystonic posturing, suggesting that the basal ganglia are involved in this clinical behavior.¹⁰¹

Impairment of consciousness, which is a feature of bilateral limbic seizures and the hallmark of CPS, is difficult to define, and can also be difficult to document in individual patients. Consciousness is a multidimensional phenomenon. The ILAE has recommended referring to cognition rather than consciousness, and has defined dyscognitive seizures as involving at least two of five components of cognition: perception, attention, emotion, memory, and executive function.¹¹ Consequently, it is unlikely that a single pathophysiologic mechanism or anatomic substrate is responsible for dyscognitive ictal manifestations. Unilateral electrical stimulation of human limbic structures can produce signs and symptoms of limbic seizures that typically occur in clear consciousness, but impairment of consciousness does not occur in the absence of contralateral discharge.^59,66 Although this suggests that at least some dyscognitive features of limbic seizures require bilateral hippocampal or other limbic involvement in the epileptic discharge, it does not rule out the possibility that certain unilateral subcortical structures that receive preferential input from the limbic system mediate other aspects of dyscognitive behavior.¹⁴⁹ Accurate descriptions of the various types of cognitive dysfunction that occur during limbic seizures, and their neurobiologic basis, cannot be elucidated from research with animal models of MTLE and will require further studies in patients.

The evolution of epileptogenesis in the limbic system, as well as propagation patterns during individual limbic seizures, have been studied in detail in animals using procedures such as amygdala kindling^22,23,60,106 or chronic models of hippocampal sclerosis following status epilepticus induced by neurotoxins or electrical stimulation.¹⁰⁵ In these models, the limbic ictal manifestations often secondarily generalize, and in amygdala-kindled animals, once the initial stages of limbic ictal manifestations progress to involve the generalized stages of rearing and falling, subsequent stimulation almost always results in the fully developed kindled seizure.¹⁰⁶ In contrast, limbic seizures only relatively rarely secondarily generalize in patients, despite the fact that ictal discharges often commonly become bilateral. This could possibly be attributed to the fact that patients are treated with antiepileptic drugs (AEDs), which are much more effective at preventing secondarily generalized seizures than they are at preventing limbic ictal events, or inherent protective mechanisms could account for failure of propagation to brainstem structures responsible for more generalized ictal manifestations. The latter interpretation is consistent with the observation that some rat models of MTLE with HS (that are not treated with AEDs) continue to have frequent localized limbic seizures that only rarely progress to involve more generalized ictal semiology.¹⁵ When limbic seizures secondarily generalize, the motor manifestations are usually asymmetrical.⁹³ Increasing evidence suggests that the responsible subcortical mechanisms are not exactly the same as those that give rise to primarily generalized tonic–clonic seizures, which are invariably symmetrical.⁷⁶

Recent studies provide converging evidence that the clinical expression of limbic seizures involves large neural networks with both inhibitory and excitatory functions.^{5,12,34,35,135} Understanding these networks might provide new targets for AEDs.⁴⁸ Also, careful network analysis might explain some of the reasons for surgical failures in limbic epilepsy.¹¹⁵ This, in turn, could lead to more effective surgical approaches.

A final consideration with respect to the ictal network responsible for the manifestations of limbic seizures concerns the unlikelihood that this will be clearly defined as a group of interrelated structures that are necessary and sufficient for limbic seizures to occur. The central nervous system is extremely redundant, and decades of animal research with limbic kindling has revealed that no forebrain lesions, including destruction of the amygdala itself,⁷⁸ can completely prevent the progression of limbic epileptogenesis.^22,23,97 Although these studies have identified specific structures that undoubtedly are important in limbic epileptogenesis, because specific lesions can delay the progression of ictal manifestations, eventually alternate pathways are found and kindling proceeds to completion. Given the fact that the human nervous system is much more complex than that of the lower animals used for kindling experiments, it is reasonable to assume that there are multiple overlapping neuronal networks capable of mediating the clinical behaviors characteristic of limbic ictal events. On the other hand, animal experiments suggest that some areas of the limbic system, apart from the hippocampus, are exquisitely epileptogenic and may be more important than others in promoting limbic epileptogenesis and epileptogenicity, such as the prepiriform, perirhinal, and entorhinal cortices.^53,67,98 In fact, a postoperative MRI study of patients who underwent amygdalohippocampectomy for MTLE demonstrated that seizure freedom is related specifically to the amount of entorhinal cortex, not hippocampus, removed.¹¹⁹

Clinical features of limbic seizures encompass a rich and diverse spectrum of signs and symptoms. Some of these reflect the region of seizure origin, some reflect spread patterns, whereas others are nonspecific. Although many of the clinical characteristics of limbic seizures have been recognized for over 100 years,^63,74,130 the advent of closed-circuit video and EEG monitoring has allowed repeated detailed studies of many seizures from large numbers of patients.^61,65

This section briefly describes the classical clinical features of limbic seizures that occur with MTLE. Limbic seizures that result from propagation of epileptic discharges originating in neocortical epileptogenic regions outside the mesial temporal lobe area often, but not always, begin with signs or symptoms that reflect the unique functioning of the cortex of origin. These variations are described in detail in Chapter 246.

Early attempts to subclassify limbic seizures based on anatomic substrates have not stood the test of time,^29,137 although they did help direct attention to different clinical manifestations. Nevertheless, it is now recognized that most limbic seizures begin in mesial temporal structures.^33,138,150 This observation, combined with specific clinical features and a distinctive hippocampal pathology, serves to define a syndrome of MTLE with HS,¹⁴⁰ the subject of Chapter 247. A generic description of a typical mesial temporal lobe seizure (MTLS) is included here.