Introduction

Limbic encephalitis (LE) is an immune-mediated neurologic disease that typically presents with memory deficits, psychiatric symptoms, and/or seizures [1]. Common etiologies include herpes simplex virus encephalitis (HSV-E) and autoimmune limbic encephalitis (ALE) [2]. On brain magnetic resonance imaging (MRI) medial temporal lobe T2-hyperintensity is classically seen in LE but is not pathognomonic of this disease [1, 3, 4]. Peri-ictal signal abnormality can have a similar radiographic appearance and cause diagnostic confusion [5, 6]. Medial temporal lobe diffusion restriction has been reported in patients with a recent seizure, indicating that diffusion-weighted imaging (DWI) may help distinguish between peri-ictal signal abnormality and LE [7,8,9]. Among patients with medial temporal lobe T2-hyperintensity, we thus sought to identify patients with recent seizure activity as well as patients with LE to look for diffusion restriction patterns that could help differentiate between these two entities.

Methods

We searched radiology reports from 1999 to 2019 in the Mayo Clinic Rochester Electronic Medical Record (EMR) using the Advanced Cohort Explorer search engine to identify patients with medial temporal lobe T2-hyperintensity attributed to recent seizure. An established non-encephalitic cause of seizure was required for inclusion. The following search terms were queried in brain MRI reports: ‘limbic AND seizure’, ‘seizure-related’, ‘seizure related’, ‘post-ictal’, ‘postictal’, ‘status epilepticus’, ‘mesial temporal T2-hyperintensity’, ‘mesial temporal T2 hyperintensity’, ‘medial temporal T2-hyperintensity’, ‘medial temporal T2 hyperintensity’, ‘limbic encephalitis’, or ‘herpes encephalitis’, which returned 3022 unique patients. We then proceeded as per the Flow Diagram. Patients with a clear cause of medial temporal lobe T2-hyperintensity other than peri-ictal signal abnormality (e.g. stroke, tumor) were excluded. Remaining patients were queried to identify those with an electroencephalogram (EEG) performed within 24 h of MRI, which was used as a surrogate marker of high suspicion for seizure. Of these patients, radiology reports were again reviewed to identify those with medial temporal lobe T2-hyperintensity deemed to be of likely peri-ictal origin (e.g. seizure listed in MRI indication, resolution of abnormality on short-term repeat MRI); mention of a seizure by the reporting radiologist, however, was not required for inclusion. The EMR of these patients were then reviewed to identify those with medial temporal lobe T2-hyperintensity that was definitely peri-ictal in origin. This was defined as (1) clinical and/or electrographic seizure documented within 48 h of MRI; (2) non-LE etiology of seizure identified; and (3) no other cause of medial temporal lobe T2-hyperintensity. Although EEG performed within 24 h of MRI was a criterion for inclusion, clinical and/or electrographic seizure documentation was required within 48 h for medial temporal lobe T2-hyperintensity to be classified as definitely peri-ictal in origin. This decision was made to increase the study sample size while ensuring a close temporal relationship between brain MRI and seizure activity. The DWI sequences of patients meeting these criteria were then reviewed to identify diffusion restriction patterns potentially unique to recent seizure activity.

figure a

We next evaluated whether any identified peri-ictal diffusion restriction patterns were observed in patients with LE. Among our patients with medial temporal lobe T2-hyperintensity we identified 31 patients with HSV-E defined by cerebrospinal fluid (CSF) HSV polymerase chain reaction positivity, and 26 patients with ALE defined by the positivity of a neural antibody in serum and/or CSF (e.g. LGI1, CASPR2, AMPAR, GABA(B)R, ANNA-1) via testing at the Mayo Clinic as previously described [10, 11]. In this LE comparison cohort, brain MRI at presentation and up to 30 days thereafter was reviewed to look for diffusion restriction patterns we had identified in our non-LE seizure cohort.

In all suspected cases, the presence of any diffusion restriction pattern was confirmed by a board-certified radiologist with a certificate of added qualification in neuroradiology who was blinded to the diagnosis (G.B.L).

Results

We identified 10 patients with peri-ictal medial temporal lobe T2-hyperintensity. In all patients, the finding was unilateral, with evidence of ipsilateral focal-onset seizures. Nine of 10 patients (90%) had medial temporal lobe diffusion restriction; four had gyriform hippocampal diffusion restriction (“Pattern 1”, Fig. 1), three had diffuse hippocampal diffusion restriction that spared the most medial temporal lobe structures (“Pattern 2”, Fig. 2), and two had both gyriform and diffuse hippocampal diffusion restriction patterns (Table 1). The median patient age was 62 years (range 2–76 years) and 3/9 (33%) were female. All patients had prolonged and/or recurrent seizures concerning for status epilepticus (SE). Repeat brain MRI was performed greater than one month after the acute presentation in 6/9 (66%) in patients with diffusion restriction. All six demonstrated hippocampal atrophy, in addition to the resolution of diffusion restriction as would be expected with signal abnormality related to acute seizure activity.

Fig. 1
figure 1

Gyriform hippocampal diffusion restriction in patients with recent seizure (“Pattern 1”). Axial T2-weighted FLAIR imaging shows left medial temporal lobe T2-hyperintensity in a patient with recent seizure activity (A1, A2, red arrows). In the same patient, DWI shows gyriform left hippocampal diffusion hyperintensity (A3, A4, red arrows). A similar diffusion restriction pattern is seen in a second patient (B1, B2) and third patient (C1, C2) with seizure activity; apparent diffusion coefficient (ADC) map of the third patient shows representative corresponding hypointensity in keeping with true diffusion restriction (C3, C4, red arrows)

Full size image
Fig. 2
figure 2

Diffuse hippocampal diffusion restriction in patients with recent seizure (“Pattern 2”). Axial T2-weighted FLAIR imaging shows left medial temporal lobe hyperintensity in a patient with recent seizure activity (A1, A2, red arrows). In the same patient, DWI shows diffuse left hippocampal diffusion hyperintensity (A3, A4, red arrows) with sparing of the most medial temporal lobe structures (A3, blue arrow). A similar diffusion restriction pattern is seen in a second patient (B1, B2) and third patient (C1, C2) with seizure activity; ADC map of the third patient shows representative corresponding hypointensity in keeping with true diffusion restriction (C3, C4, red arrows)

Full size image
Table 1 Characteristics of patients with peri-ictal diffusion restriction patterns
Full size table

In comparison, only 5/57 patients with LE (9%) had one these diffusion restriction patterns (P < 0.0001, Fisher’s Exact test). Of the 31 with HSV-E, four (13%) had “Pattern 2” diffusion restriction; one presented with confusion and developed tonic–clonic SE less than 24 h after MRI, one presented with confusion and was in focal-onset SE on EEG less than 24 h after MRI, and two had a high clinical concern for focal-onset SE (one presented with confusion followed by right facial twitching and automatisms less than 48 h after MRI, the other presented with left head deviation and shaking movements followed by a persistent decreased level of consciousness less than 48 h before MRI). All four of these patients had repeat brain MRI performed within one week of acute presentation, and progressive diffusion restriction outside the medial temporal lobe involving typical regions of HSV-E (i.e. anterior temporal lobe, insular cortex and/or cingulate cortex) was observed in all [2]. One of four patients had a repeat brain MRI that was performed greater than one month after acute presentation, which showed resolution of diffusion restriction and temporo-insular encephalomalacia. Of the 26 with antibody-positive ALE (17 LGI1, 5 ANNA-1, 3 CASPR2, 1 AMPAR), only one (4%) with LGI1-antibody encephalitis had “Pattern 2” diffusion restriction. This patient presented with confusion and initial EEG showed only slowing, but repeat EEG for persistent cognitive difficulties 2 weeks later showed focal electrographic seizures. Repeat brain MRI performed within one week of the acute presentation showed additional subtle diffusion restriction of the caudate, which has been previously described in LGI1-antibody encephalitis [12]. Repeat brain MRI performed greater than one month after acute presentation showed resolution of diffusion restriction, as well as hippocampal atrophy and caudate encephalomalacia.

Discussion

Among patients with medial temporal lobe T2-hyperintensity, we describe two patterns of diffusion restriction that can help differentiate peri-ictal signal abnormality from LE. Although LE may be investigated as a substrate of seizure activity in this setting, other etiologies should be considered if supportive evidence of LE is lacking.

Medial temporal lobe diffusion restriction has been reported previously with recent seizure activity, and presumably reflects transient seizure-induced vasogenic and cytotoxic edema [5, 7]. However, diffusion restriction in this region has also been described in infectious and autoimmune encephalitis [13, 14], indicating that the simple presence of diffusion restriction cannot distinguish seizure-related change from LE. Furthermore, patients with LE are at high risk for seizures, making it difficult to determine whether diffusion restriction reported in the previous series was related to LE or recent seizure activity [13]. For this reason, a systematic investigation of diffusion restriction patterns that can help differentiate between these two entities is needed.

There are several limitations to this retrospective study. Due to the stringency of criteria that needed to be met before classifying medial temporal lobe T2-hyperintensity as peri-ictal in origin, the number of patients who were ultimately included in this cohort was small. These highly selective criteria, however, served to minimize the possibility of factors other than recent seizure contributing to the medial temporal lobe T2-hyperintensity observed. In particular, the requirement of an established cause of seizure other than LE minimized the likelihood of antibody-negative ALE in this cohort, permitting identification of diffusion restriction patterns attributable to recent seizure activity. Furthermore, because our search strategy was designed to identify patients with a high likelihood of peri-ictal signal abnormality, clinical sensitivity and specificity of the diffusion restriction patterns we identified for recent seizure activity in all patients with medial temporal lobe T2-hyperintensity could not be determined. Nonetheless, the diffusion restriction patterns we identified were reproducibly seen in patients with peri-ictal medial temporal lobe T2-hyperintensity and were also associated with seizures in patients with medial temporal lobe T2-hyperintensity and LE.

Recognition of these diffusion restriction patterns in patients with medial temporal lobe T2-hyperintensity should prompt consideration of seizure rather than LE as the cause of the signal abnormality. Even in patients with LE, the presence of these patterns should raise concern for seizure.