Historical note and nomenclature

Hippocampal atrophy is the hallmark of hippocampal sclerosis, the most frequent neuropathological finding in patients with pharmacoresistant mesial temporal lobe epilepsy submitted to surgical treatment (84; 12; 24). The association with chronic seizures was first described in 1825 by Bouchet and Cazauvielh, who identified an increased consistency of hippocampus (hardening or sclerosis) in postmortem specimens of patients with epilepsy, thereby originating the terminology of hippocampal sclerosis or Ammon horn sclerosis (15). In 1880, Sommer acknowledged the importance of this type of lesion in patients with epilepsy and described the selective neuronal loss in the “Sommer sector,” now recognized as cornu ammonis area 1 CA1 (84).

According to the proposed terminology by the International League Against Epilepsy (ILAE), mesial temporal lobe epilepsy with hippocampal sclerosis is classified as one of the distinctive constellations within the section of electroclinical syndromes and other epilepsies (07). Hippocampal atrophy as a component of mesial temporal lobe epilepsy is also part of the mesial temporal lobe epilepsy complex observed in familial mesial temporal lobe epilepsy and can also be associated with malformation of cortical development, or, more specifically, focal cortical dysplasia type IIIa (focal cortical dysplasia associated with hippocampal sclerosis) (14). In addition, hippocampal atrophy can occur in the context of dual pathology, which is described as an “extrahippocampal lesion plus hippocampal atrophy” (22; 60; 14).

Clinical manifestations

Presentation and course

The typical history of mesial temporal lobe epilepsy with hippocampal sclerosis describes the occurrence of an initial precipitating injury, such as prolonged febrile seizures, CNS infection, and trauma, followed by a latent period until the onset of habitual seizures (64). However, the initial precipitating injury is not frequently identified (24). Particular attention has been directed to the well-known association between prolonged febrile seizures (and febrile status epilepticus) in early infancy and the development of mesial temporal sclerosis (20). However, the exact causal relationship between febrile seizures, temporal lobe epilepsy, and hippocampal sclerosis has not yet been totally elucidated. Advances in clinical and molecular genetic studies strongly suggest a role of genetic factors, implying a causal role of febrile status epilepticus in the development of temporal lobe epilepsy (72).

The first ictal event may be either a focal seizure with impaired awareness or a focal to bilateral tonic-clonic seizure (38). Usually, before the focal impaired awareness seizures, patients present a typical aura characterized by an indeterminate visceral sensation, described as a rising epigastric sensation, nausea, or “butterflies.” It can be associated with fear, autonomic (such as tachycardia, pallor), or other psychic (eg, déjà vu, jamais vu) symptoms. Some patients may also present olfactory and gustatory sensations (24; 72). The occurrence of isolated auras without focal impaired awareness seizures is not uncommon, even years preceding the habitual seizures (55; 54; 06).

The focal impaired awareness seizures characteristically involve impairment of consciousness or awareness and usually initiate with motionless staring that may or may not be accompanied by oroalimentary automatisms (eg, chewing, lip smacking) (07; 06). Concurrent to the impairment of awareness, there may be additional lateralization signs, such as contralateral dystonic posturing or verbal automatisms (when seizures arise from the nondominant hemisphere). A postictal confusional state or aphasia (if the seizure arises from the language-dominant hemisphere) is frequent (24).

Mesial temporal lobe epilepsy with hippocampal sclerosis is highly associated with pharmacoresistant seizures; however, there are patients who respond well to antiseizure medication (55; 58). Patients with mesial temporal lobe epilepsy and hippocampal atrophy who have been seizure-free for at least 24 months with or without medication, and who are not surgical candidates, have been recognized more frequently (02). Some of these patients belong to families with mesial temporal lobe epilepsy (55). Most patients with familial mesial temporal lobe epilepsy have few seizures during their lifetimes or respond well to antiseizure medication. Over time, they persist with the same clinical status (ie, they do not develop pharmacoresistant epilepsy), as observed in a longitudinal study of 17 families with long-term follow-up (69). There are also patients with sporadic mesial temporal lobe epilepsy with late-onset seizures who respond well to antiseizure medication (58).

Patients with pharmacoresistant mesial temporal lobe epilepsy associated with hippocampal atrophy have significantly more social, psychiatric, and cognitive dysfunction than those with well-controlled seizures and, consequently, more stigma and difficulties for employment and schooling (56; 71). These difficulties are, in part, associated with recurrent seizures and, in part, related to cognitive dysfunction, which typically affects memory (56). Verbal memory is predominantly affected when the hippocampal atrophy occurs in the hemisphere dominant for language; on the contrary, when the hippocampal atrophy is in the nondominant hemisphere, deficits in visual memory are more severe (01). In addition, these subjects present other behavioral comorbidities such as depression and psychosis, which may contribute to social problems (77; 51; 71). Also, temporal lobe epilepsy patients with left-sided hippocampal atrophy have worse brain dysfunction as compared to those with right-sided hippocampal atrophy and those without atrophy (74; 94).

Especially in pharmacoresistant patients, accidents may occur involving burns, bone and teeth fractures, as well as head trauma due to frequent falls (30). There is also an increased risk of sudden death in these patients (05).

Pathophysiology

Hippocampal atrophy associated with hippocampal sclerosis in epilepsy is a network disorder, functionally and structurally (10; 93; 31; 74; 94). Hippocampal sclerosis is characterized by a selective pattern of neuronal loss within the subfields that distinguish it from other types of hippocampal atrophy not related to epilepsy, such as in Alzheimer disease or normal aging. In addition to the segmental neuronal cell loss, there is usually astrogliosis, mossy fiber sprouting, and granule cell dispersion (24; 65). Over the years, there have been many classifications of hippocampal sclerosis proposed, and in 2013, a neuropathology taskforce was designated by the ILAE to incorporate the previous schemes into a consensus (13). The histopathological characterization includes three types of hippocampal sclerosis with one additional term: “Gliosis only, no hippocampal sclerosis” (13).

Hippocampal sclerosis ILAE type 1. This type is the most frequent pattern identified in the surgical series (classic and total forms of hippocampal sclerosis, 60% to 80% of the cases). There is approximately 80% loss of neurons in the CA1 section, associated with loss of neurons in CA2 (30% to 50%), CA3 (30% to 90%), and CA4 (40% to 90%). There is also loss of granule cell of dentate gyrus (approximately 50% to 60%).

Hippocampal sclerosis ILAE type 2. This type represents a rare pattern of hippocampal sclerosis (5% to 10%). In this pattern, the entire CA1 section is affected, with mild neuronal loss in the other subfields.

Hippocampal sclerosis ILAE type 3. This type is also an infrequent pattern of hippocampal sclerosis, occurring in only 4% to 7% of the surgical cases. There is marked loss in CA4 (approximately 50%) and dentate gyrus (more than 30% of granular neurons), with moderate neuronal loss in the remaining subfields. It has been associated with dual pathologies and limbic encephalitis.

Gliosis only, no hippocampal sclerosis. This type is present in nearly 20% of surgical specimens and shows reactive gliosis without significant neuronal loss.

The diagnosis of dual pathology is applied when hippocampal sclerosis occurs in the presence of other pathologies outside the mesial temporal lobe (neoplasm, vascular lesions, glial scar, limbic/Rasmussen encephalitis, or malformation of cortical development) (14; 24).

The mechanisms underlying the development of hippocampal sclerosis remain undetermined. Hippocampal sclerosis most likely has different causes in different individuals and may result from complex interactions among genetic and environmental factors (24).

Studies have shown that prolonged and focal febrile seizures can produce acute hippocampal injury that evolves to hippocampal atrophy and that complex febrile seizures can originate in the temporal lobes in some children (66). Although there is a high incidence of complex febrile seizures among patients with mesial temporal sclerosis in retrospective studies, it is still not clear whether complex febrile seizures are an epiphenomenon or a causative factor, or both (63).

Genetic predisposition appears to be an important causal factor in patients with hippocampal sclerosis and antecedent prolonged febrile seizures (08; 52). Clinical and molecular genetic studies show that there is some specificity in the types of epilepsy that follow febrile seizures, rather than febrile seizures being a nonspecific marker of a lowered seizure threshold. The relationship between febrile seizures and later development of epilepsy appears to have a strong genetic background (08).

Evidence from experimental and clinical studies also provides data suggesting the importance of innate immunity in the etiology and pathogenesis of mesial temporal lobe epilepsy. The presence of inflammatory processes, indicated, for example, by the presence of highly upregulated chemokines genes in chronic mesial temporal lobe epilepsy (86) may directly and indirectly affect neuronal excitability.

Studies of autoimmune epilepsy and limbic encephalitis have described pharmacoresistant seizures, cognitive deficits, and presence of abnormalities in the mesial temporal region, including hippocampal atrophy (87; 19; 37). Autoimmune limbic encephalitis has been associated with a variety of pathogenic antibodies targeting neuronal antigens, including LGI1 (leucine-rich, glioma inactivated 1), NMDA receptor (N-methyl-D_aspartate), as well as onconeural antibodies such as Hu and Ma2Ta and others (59; 40).

The relationship between autoimmune epilepsy, limbic encephalitis, and hippocampal atrophy remains unclear and requires further longitudinal studies. Whether some subjects with temporal lobe epilepsy and hippocampal atrophy represent “mild forms of limbic encephalitis” is debatable and needs further investigation (40). According to Gaspard, some clinical features should raise the awareness of possible autoimmune basis for epilepsy, including: onset with status epilepticus, early pharmacoresistance, rapid cognitive decline, onset after 30 years, mesial temporal inflammatory findings on brain MRI, and bilateral mesial temporal findings on brain MRI (40). Future studies may clarify the impact of the autoimmune process on development of hippocampal atrophy and perhaps allow new treatment strategies to avoid pharmacoresistant seizures.

Other brain injuries and infections may be related to the underlying causes of hippocampal atrophy. Chronic autoimmune inflammation after viral encephalitis, such as herpes virus, could also be a potential underlying cause of hippocampal atrophy, in addition to the acute damage caused by the viral infection (76). There is a high frequency of hippocampal atrophy in patients with calcified neurocysticercosis and pharmacoresistant seizures, suggesting a possible association between these two conditions (81; 47; 78; 79). Some mesial temporal lobe epilepsies with hippocampal atrophy, usually bilateral, appear to result from traumatic brain injuries (34).

Differential diagnosis

The typical seizures in mesial temporal lobe epilepsy with hippocampal sclerosis seizures cannot be clinically distinguished from seizures secondary to other lesions in mesial temporal area (eg, tumors, vascular malformations, or malformations of cortical development), nor can it be differentiated from seizures arising from the same region in the absence of structural abnormalities detected on MRI. It is always necessary to evaluate the signs and symptoms along with the results from diagnostic tests, including MRI and EEG (24; 25).

Diagnostic workup

Neuropsychological tests. Neuropsychological evaluation commonly demonstrates memory dysfunction, which is material-specific according to the hemisphere involved and is related to the degree of hippocampal atrophy on MRI and cell loss on postoperative histopathology. Verbal memory is mostly affected with left-sided hippocampal atrophy, whereas visuospatial memory is more affected with right-sided hippocampal atrophy (50).

EEG. Interictal EEG findings in patients with mesial temporal lobe epilepsy and hippocampal atrophy typically include unilateral or bilaterally independent mesial temporal spikes, best seen with basal (sphenoidal, inferior temporal) derivations. In patients with MRI-defined unilateral hippocampal sclerosis, the interictal epileptiform discharges are predominantly or exclusively ipsilateral to the hippocampal atrophy, even after decades of ongoing seizures (17). Contralateral interictal epileptiform discharges appear to reflect early widespread abnormalities and not epilepsy progression. Temporal intermittent rhythmic delta activity appears to have a localizing value of the epileptogenic zone in mesial temporal lobe epilepsy, unlike intermittent rhythmic delta activity in other brain regions (39).

Ictal EEG recordings usually reveal ictal onset consisting of regular, well-lateralized, rhythmic 5- to 9-Hz activity in one anterior-mid and inferomesial temporal region before the first clinical manifestations or within 30 seconds (delayed focal onset), with or without contralateral propagation. However, artifacts due to oromasticatory movements and other automatisms early in the clinical manifestations frequently obscure the ictal scalp EEG recordings (23).

Ictal discharges may be confined to the medial temporal structures for a few seconds or longer, without evident EEG changes on scalp recordings. This may be followed by a fast propagation of seizure discharges to the ipsilateral or contralateral temporal neocortex, or it may propagate to both hemispheres in a diffuse fashion. In these circumstances, the scalp EEG record may miss the initial (truly localizing) seizure discharges. Thus, ictal scalp EEG recordings may have limited localizing value when the first clinical manifestations clearly precede the first EEG changes. This false localization phenomenon may be associated with severe hippocampal damage (67).

Although diagnosis of mesial temporal lobe epilepsy with hippocampal sclerosis and identification of the side of ictal onset for surgical therapy is now possible in most patients using noninvasive investigation (03), additional long-term monitoring with intracranial electrodes may be indicated when the side of mesial temporal ictal onset is unclear or there remains a possibility of neocortical ictal onset (48; 35). Most centers utilize depth electrodes for this purpose, but subdural strips or grids and foramen ovale electrodes can also be used (23).

Neuroimaging. When there is a clinical suspicion of mesial temporal lobe epilepsy, MRI can provide a noninvasive and highly accurate diagnostic method to detect hippocampal atrophy and other signs of hippocampal sclerosis in vivo (24). The optimization of acquisition protocol is necessary for adequate evaluation of hippocampal morphology, including thin high-resolution T2-weighted coronal slices (3 mm maximum) obtained on a plane perpendicular to the long axis of the hippocampus; high-resolution 1 mm volumetric (3D) T1-weighted images; and coronal, or preferentially, 3D-FLAIR (fluid attenuation inversion recovery) images (09). The visual inspection of MRI may provide clear information when the asymmetry of hippocampi is evident, facilitating the identification of hippocampal atrophy. Typically, atrophy of the hippocampus is associated with increased T2-weighted signal (24).

Studies with proton magnetic resonance spectroscopy (¹H-MRS) show associated reduction of neuronal marker N-acetylaspartate in mesial temporal lobe epilepsy (even in subjects without clear hippocampal sclerosis) (21; 74). The quantification of T2 relaxometry may provide additional information about the lateralization as hyperintense T2 signal in the hippocampus is associated with hippocampal sclerosis and concordant with EEGs (28; 29).

However, the visual analysis may not offer straightforward results when there is bilateral atrophy or mild unilateral hippocampal atrophy. As the identification of atrophic hippocampus may be crucial for surgical planning, additional tools can be used to detect more subtle abnormalities that can define the lateralization of atrophy. Volumetry of the hippocampus can provide very accurate information about lateralization of the temporal lobe epilepsy, which can be used for surgical treatment (21). Automatic models to obtain hippocampal volume have been developed and compared in order to facilitate the identification of morphometric abnormalities (70; 62). Automated imaging assessment tools can assist clinical radiology reporting of hippocampal sclerosis, improving detection accuracy in clinical practice (42; 57). Machine learning algorithms using multimodal MRI have been promising in lateralizing hippocampal pathology in patients with temporal lobe epilepsy and hippocampal sclerosis and may have clinical applications in the near future (18; 41).

Other functional tests such as [18F] fluorodeoxyglucose (FDG)-PET and ictal SPECT may help to lateralize the seizure origin, especially when EEGs show bilateral epileptiform activity (44; 91; 24). The temporal lobe with hippocampal atrophy is usually hypometabolic on interictal FDG-PET, with an area that involves the mesial structures and other parts of the temporal lobe, which is helpful and reliable in localization for surgical treatment (24; 25). Although ictal and early postictal SPECT scans can be helpful in localizing the epileptogenic focus in patients with mesial temporal lobe epilepsy, it is logistically difficult and is not routinely performed in most centers (88).

Syndromes and diseases in which the seizure type occurs

Adolescent to adult

Prognosis and complications

According to a hospital-based survey, temporal lobe epilepsy was the most refractory focal epilepsy (80). Temporal lobe epilepsy with hippocampal sclerosis was associated with a very low seizure freedom rate (11% of patients). Despite the lack of population-based epidemiological studies to evaluate the outcome of temporal lobe epilepsy, there are some patients with mesial temporal lobe epilepsy and hippocampal atrophy that do not progress to refractory seizures and respond well to medication or remit over time (55; 69).

For those who fail at least two appropriate antiseizure medication regiments, seizure control remains poor, despite the innumerous attempts at polytherapy schemes (53). These patients usually take high doses of antiseizure medication and still present with frequent focal impaired awareness seizures, usually several per week or per month. Therefore, the occurrence of side effects is common, especially associated with high doses of polytherapy. Some examples are weight gain, hair loss and tremor (valproic acid), gingival hyperplasia (phenytoin), and language and memory disturbances (topiramate), among others (92; 26).

There is imaging and clinical evidence of progression of damage in patients with temporal lobe epilepsy associated with hippocampal atrophy, which appears to be only partially related to seizures (02; 83; 61). Cognitive decline, worsening of memory, and psychiatric comorbidities are among the most relevant clinical complains of these patients over the years (56; 71; 83).

Management

Clinical treatment. The initial treatment should aim at maximum tolerated doses of antiseizure medications, taking into account that some patients may respond well to treatment despite the presence of hippocampal atrophy. However, it is important to recognize pharmacoresistant patients without delay (85). These patients should be referred to comprehensive epilepsy centers for evaluation of possible surgical treatment to prevent further cognitive dysfunction as well as death (Jette and Engel 2016). There is Class I evidence for better efficacy of surgery compared to continuing with further antiseizure medication trials in patients with mesial temporal lobe epilepsy with hippocampal sclerosis who failed treatment with two or more AEDs (89; 36).

The initial treatment should be a single antiseizure medication at a low dose gradually incremented over weeks to establish an effective and tolerable regimen. Further dose adjustments can be made according to the clinical response (43). The choice of antiseizure medication needs to be assessed individually and consider several factors, including the adverse effect profile of each medication and the potential teratogenic effect on women, age, and comorbidities.

Second- and third-generation antiseizure medications are less likely to cause pharmacokinetic interactions than older antiseizure medications (73). Some second-generation antiseizure medications have advantages in tolerability and safety, particularly in treating older patients and women of childbearing potential (eg, lamotrigine). Unfortunately, none of the newer antiseizure medications have more efficacy than the older antiseizure medications (73).

A study comparing the effectiveness of antiseizure medications in patients with mesial temporal lobe epilepsy with hippocampal sclerosis showed that seizure improvement was highest for the first antiseizure medications prescribed and lowest for subsequent antiseizure medications (04). Seizure‐freedom rates were highest with carbamazepine, levetiracetam, and oxcarbazepine. The highest 12‐month retention rates were with carbamazepine (85.9%), valproate (85%), and clobazam (79%) (04). The highest adverse reaction rates were with oxcarbazepine (35.7%), topiramate (30.9%), and pregabalin (27.4%), and the lowest rates were with clobazam (6.5%), gabapentin (8.9%), and lamotrigine (16.6%).

Unfortunately, a significant proportion of patients with mesial temporal lobe epilepsy with hippocampal sclerosis will eventually present with refractory seizures and require polytherapy. In general, polytherapy increases the risk of undesired side effects and teratogenicity (45; 90; 16; 04; 73). The ideal combination should provide “supra-additive efficacy with infra-additive toxicity” (16). However, this is not always achieved and usually yields some degree of undesired side effects. Robust evidence to guide when and how to combine antiseizure medications is still lacking (82; 73). Therefore, the decision should consider both comedication and comorbidities to optimize the best options individually. Some examples of synergism are the combination of lamotrigine and valproate, carbamazepine with clobazam, and lacosamide with levetiracetam (75; 68; 27).

Despite the absence of evidence-based guidelines to recommend appropriate immunotherapy in patients with limbic encephalitis and hippocampal atrophy, current alternatives (based mostly on observational evidence) include intravenous steroids, intravenous immunoglobulin, plasma exchange, rituximab, and cyclophosphamide (40). The combination of immunotherapy and antiseizure medications provides relatively good control of seizures (87), although some subjects develop hippocampal atrophy and persist with pharmacoresistant seizures, requiring surgical treatment (19).

Surgical treatment. The removal of mesial temporal lobe structures has been indicated for pharmacoresistant patients (89). Different surgical approaches have been proposed, including more selective amygdalohippocampectomy and standard temporal lobectomy (46; 33). There are two class I studies confirming the superiority of surgical treatment over prolonged medical treatment for patients who had inadequate responses to antiseizure medications (89; 36). In one of the studies, 23 patients continued antiseizure medications and 15 underwent anteromesial temporal resection with antiseizure medication treatment. After two years of follow-up, 73% of operated patients were seizure-free, compared to none of the medical group (36).

In a study of 213 patients with a neuropathological diagnosis of hippocampal sclerosis and a minimum follow-up of two years, the authors divided the subjects according to ILAE hippocampal sclerosis type 1 or 2 and further classified them into three groups: isolated hippocampal sclerosis, hippocampal sclerosis associated with focal cortical dysplasia, or hippocampal sclerosis associated with other lesions (32). Regardless of the hippocampal sclerosis type and associated pathology, they observed greater than 80% of Engel class I for both short- and long-term outcomes. Patients with hippocampal sclerosis type 2 (cell loss predominantly in CA1) presented a better long-term outcome than those with hippocampal sclerosis type 1 (cell loss in all sectors of CA). The presence of focal cortical dysplasia was associated with a worse outcome (regardless of hippocampal sclerosis type), whereas a shorter duration of epilepsy correlated with seizure freedom after surgery (32).

Given the evidence, efforts should be directed to reduce the time to surgical referral, offering patients higher chances of earlier seizure control and improvement in quality of life and avoidance of disability (36).