Seizure: European Journal of Epilepsy 92 (2021) 166–173

Contents lists available at ScienceDirect

Seizure: European Journal of Epilepsy

journal homepage: www.elsevier.com/locate/seizure

Review

Seizure-induced reversible MRI abnormalities in status epilepticus: A

systematic review
Frederick P. Mariajoseph a, Parveen Sagar b, Subramanian Muthusamy b, Shalini Amukotuwa c,
Udaya Seneviratne a, b, d, *
a
School of Clinical Sciences at Monash Health, Department of Medicine, Monash University, Melbourne, Victoria, Australia
b
Department of Neurology, Monash Medical Centre, Clayton, Melbourne, Australia
c
Monash Imaging, Monash Health, Clayton, Melbourne, Australia
d
Department of Medicine, St. Vincent’s Hospital, University of Melbourne, Melbourne, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: In the context of status epilepticus (SE), seizure-induced reversible MRI abnormalities (SRMA) can be difficult to
Epilepsy differentiate from epileptogenic pathologies. To identify patterns and characteristics of SRMA, we conducted a
Transient systematic review in accordance with the Preferred Items Reporting for Systematic Reviews and Meta-Analyses
EEG
(PRISMA) guidelines. We included publications describing patients (a) presenting with status epilepticus, (b)
Imaging
Periictal
exhibiting seizure-induced MRI abnormalities, (c) who demonstrated complete resolution of MRI abnormality at
Postictal follow-up, and (d) who had availability of descriptive MRI results.
A total of 49 cases from 19 publications fulfilled our eligibility criteria. Signal abnormalities were most
frequently reported on T2-weighted sequences followed by diffusion-weighted and fluid-attenuated inversion
recovery imaging. Both unilateral and bilateral SRMA were reported. Unilateral EEG abnormalities were often
associated with ipsilateral SRMA. The signal changes appeared during the ictus itself in some subjects whilst the
median time to SRMA appearance and resolution were 24 h and 96.5 days, respectively.
Based on the distribution of reversible signal alterations, we identified five ‘composite patterns’: (1) pre­
dominant cortical (with or without subcortical, leptomeningeal or thalamic involvement), (2) hippocampal (with
or without cortical, subcortical, leptomeningeal, or thalamic involvement), (3) claustrum, (4) predominant
subcortical, and (5) splenium involvement. Amongst treatment-responsive SE patients, the cortical pattern was
the most prevalent whereas hippocampal involvement was most frequently reported in refractory SE.
Cortical atrophy, hippocampal sclerosis, and cortical laminar necrosis were common long-term sequelae after
the resolution of SRMA. In this review, we highlight many limitations of the literature and discuss future di­
rections for research.

1. Introduction abnormalities (SRMA) [2,3].

Status epilepticus (SE) is a neurological emergency associated with
high rates of morbidity and mortality. The incidence of SE has been
reported to be between 9.9 and 41 per 100,000 person-years, with a
bimodal peaking in children and the elderly [1]. Ascertaining the un­
derlying aetiology of SE is an essential step of management, and mag­
netic resonance imaging (MRI) of the brain plays an important role.
Abnormalities seen on MRI do not always reflect a pathology, however,
as several studies have reported seizure-induced reversible MRI
The characteristics, natural history and clinical implications of
SRMA remain incompletely understood and often present a clinical
dilemma by mimicking epileptogenic lesions. Consequently, it can be
difficult to distinguish between SRMA and underlying pathologies such
as tumours, infections, or infarcts, which often require timely and tar­
geted management [4]. Against this backdrop, we conducted this sys­
tematic review to investigate the characteristic features of SRMA in
patients presenting with status epilepticus.

Abbreviations: ASM, antiseizure medications; SRMA, seizure-induced reversible MRI abnormalities; SE, status epilepticus.
* Corresponding author at: School of Clinical Sciences at Monash Health, Department of Medicine, Monash University, Melbourne, Victoria, Australia
E-mail address: udaya.seneviratne@monash.edu (U. Seneviratne).

https://doi.org/10.1016/j.seizure.2021.09.002
Received 14 July 2021; Received in revised form 3 September 2021; Accepted 5 September 2021
Available online 7 September 2021
1059-1311/© 2021 British Epilepsy Association. Published by Elsevier Ltd. All rights reserved.
F.P. Mariajoseph et al.
2. Methods
2.1. Search strategy
A comprehensive literature search of PubMed (1996 to December
2020), Medline (1946 to December 2020), and Embase (1947 to
December 2020) was conducted. Key search terms included variations
and synonyms of the words “seizure”, “status epilepticus”, “refractory
status epilepticus”, “super-refractory status epilepticus”, “new onset
refractory status epilepticus”, “magnetic resonance imaging”, “revers­
ible”, “induced” and “abnormality’, with Boolean operators utilised to
combine these keywords. The search was designed by two investigators
(FM & US) and performed by the first author (FM) with two investigators
(FM & PS) independently screening titles and abstracts for eligibility.
This was followed by full-text screening by the same investigators.
“Covidence”, which is an online platform was used to facilitate the
screening process (www.covidence.org). A third investigator (US)
adjudicated any discrepancies between the two investigators. In order to
identify additional publications, reference lists from selected papers
were screened by two investigators (FM & PS). This study was conducted
according to the Preferred Items Reporting for Systematic Review and
Meta-Analyses (PRISMA) guidelines [5].
2.2. Eligibility criteria
The following inclusion criteria were applied: primary research
publications with (a) patients presenting with status epilepticus, (b)
reporting of seizure-induced MRI abnormalities, (c) complete resolution
of MRI abnormality at follow-up, and (d) availability of descriptive MRI
results. Status epilepticus was defined as tonic-clonic seizures lasting for
>5 min and other seizure types >10 min in duration [6]. Reports of
single seizures, cluster seizures, provoked seizures, partial resolution of
MRI abnormalities, animal studies, grey literature, non-English litera­
ture, and review publications were excluded. Both adult and paediatric
subjects were included. All individual cases reported within publications
were evaluated according to the aforementioned inclusion and exclusion
criteria. The list of full-text articles was finalised in August 2021.
2.3. Data extraction
A data extraction spreadsheet was prepared using Microsoft Excel
with one investigator (FM) extracting the data and two investigators (PS
& SM) subsequently cross-checking this data against the source publi­
cation. A neuroradiologist (SA) reviewed all imaging data in publica­
tions and an independent investigator (US) adjudicated any
discrepancies in the tabulated data. Information collected included age,
sex, country of study, comorbidities, aetiology, treatment, anti-seizure
medications (ASMs), seizure frequency, EEG features, baseline MRI
data, duration of the index seizure, duration of epilepsy diagnosis,
seizure classification, epilepsy classification, seizure focus, response to
treatment, the interval between the index seizure and initial MRI, in­
terval between the index seizure and follow up MRI and MRI data. The
following MRI data were recorded: abnormality on a particular MRI
sequence (DWI – diffusion-weighted imaging, ADC – apparent diffusion
coefficient, FLAIR – fluid-attenuated inversion recovery, T2, T1, post-
contrast), and the brain region, laterality (unilateral and bilateral),
and anatomical localisation (cortical, subcortical, deep grey matter) of
this abnormality, and imaging sequelae (cortical atrophy, hippocampal
sclerosis, cortical laminar necrosis). Seizures were classified in accor­
dance with the International League against Epilepsy (ILAE) operational
classification of seizures [7]. Data that were not specified or reported
were recorded as “not reported”.
2.4. Quality assessment
Due to the observed variation in study designs and standards, we
167
Seizure: European Journal of Epilepsy 92 (2021) 166–173
conducted a bias and quality assessment by utilising an eight-item
questionnaire outlined by Murad et al. encompassing the domains of
selection, ascertainment, causality, and reporting.[8] Each question was
scored as a “yes” or a “no”. We assessed each case separately and cases
were assessed by two independent investigators (FM & SM) with dis­
crepancies adjudicated by a third investigator (US).
2.5. Synthesis of results
Qualitative data were summarised using tables and graphs. We also
synthesised descriptive statistics including the means and medians of
continuous variables and frequencies of discrete variables.
3. Results
3.1. Study population characteristics
Demographics and clinical characteristics of subjects are summarised
in Table 1 and further details can be found in Supplementary Table 1. Our
final analysis included a total of 19 publications [9–27]. Of the 95 cases
that these publications reported, 49 fulfilled the eligibility criteria with
10 (20%) subjects selected from single case reports and the remaining 29
(59%) being obtained from case series (Fig.1). The year of publication
ranged from 1987 to 2018. The median age of included subjects was 29
years (range: 2–79) and 21 (43%) were females. There were three cat­
egories of epilepsy and seizure presentations that we identified – (i) 29
patients (59%) had pre-existing focal epilepsy, (ii) 3 (6%) had
pre-existing generalised epilepsy, and (iii) 10 patients (20%) presented
as the first unprovoked seizure. The remaining seven subjects could not
be classified due to inadequate data.
Based on the available information in publications, we classified SE
according to the ILAE terminology [6]. Seventeen patients (35%) had
non-convulsive status epilepticus (NCSE). Thirteen patients (27%) had
convulsive status epilepticus (CSE) of focal onset and one patient (2%)
had CSE of generalised onset, whilst three (6%) presented with CSE of
unknown onset. Focal motor status epilepticus (FMSE) was reported in
two patients (4%) and one subject (2%) suffered myoclonic SE. Eight
patients (16%) presented with focal SE and details were not provided in
publications for further classification.
In regards to the seizure aetiologies reported, twelve patients (34%)
had a vascular cause [prior stroke (5), cavernoma (2), prior cortical
haemorrhage (2), cerebrovascular disease (2), cerebral microangiopathy
(1)], whilst in two patients (6%) SE was considered to be caused by
multiple sclerosis. Occipital metastasis, prior occipital resection of an
arteriovenous malformation, perinatal hypoxia, and glioblastoma mul­
tiforme were responsible for SE of one case each. Nine patients (18%)
presented as new onset refractory status epilepticus (NORSE). The
aetiology of the seizure was unspecified or unknown in 17 subjects
(49%).
With respect to ASMs before the onset of SE, two patients (4%) were
receiving phenytoin, one patient (2%) was on phenobarbital, and one
(2%) was receiving a combination of lamotrigine, phenytoin, and leve­
tiracetam. For the remaining 45 patients (92%), medications preceding
the onset of status epilepticus were not reported. Treatment to terminate
the seizure was not reported in 26 subjects (74%). Of those that did
specify, there was a wide range of different ASM combinations reported,
with phenytoin and midazolam being the most common. Thirteen sub­
jects received immunotherapy such as steroids and intravenous immu­
noglobulins. Ten patients (20%) were responsive to treatment, eight
(16%) were refractory and four (8%) were defined as super-refractory
SE. Further details regarding treatment as well as prior medications
can be found in Table 1 and Supplementary Table 1.
Upon follow-up, 16 patients (33%) were reported to be seizure-free
with a follow-up duration ranging from two to 36 months whilst
eleven patients (22%) experienced subsequent seizure episodes (follow-
up 2–12 months). Mild cognitive impairment was noted in four patients
F.P. Mariajoseph et al. Seizure: European Journal of Epilepsy 92 (2021) 166–173

Table 1 (8%) and one patient (2%) was in a comatose state at the 6-month
Summary of patient demographics, epilepsy classification and seizure follow-up. One patient (2%) died during SE, and one patient (2%) was
characteristics lost to follow-up, and for the remaining 15 cases (31%) long-term clin­
Clinical and demographic variables Number of cases ical outcomes were not reported.
(%) The EEG findings were reported with some variability. 25 patients
Median Age 29y (range: 2 to 79) (49%) had only interictal EEG data available, whilst 16 patients (33%)
Sex had only ictal EEG data. Four subjects (8%) had both interictal and ictal
Female 21 (43) EEG information, and a further four (8%) had no EEG data reported. The
Male 24 (49)
most common interictal EEG finding was focal epileptiform discharges
Unspecified 4 (8)
Country observed in 12 patients (85.7%), whilst temporal-onset epileptiform
Italy 11 (22) rhythm was the most frequently reported ictal EEG finding (9 patients).
South Korea 8 (16)
USA 6 (12) 3.2. Bias and quality assessment
Germany 6 (12)
France 4 (8)
Taiwan 4 (8) Of the 49 cases included, 38 (78%) exhibited a satisfactory selection
Portugal 3 (6) of SRMA and 39 cases (80%) appropriately excluded alternative causes.
Austria 1 (2) Follow-up was deemed acceptable in 40 patients (82%) and the quality
Hungary 1 (2)
of reporting was considered sufficient in only 24 cases (49%) reflecting
Japan 3 (6)
Switzerland 1 (2) the variation in reporting standard. Further information relating to the
UK 1 (2) bias and quality assessment of this study can be found in Supplementary
Epilepsy Classification Table 2. Of note, reports of 40 cases (82%) provided images of the
Focal epilepsy 29 (59) abnormal MRI demonstrating SRMA, whilst only 28 (57%) provided
Generalised epilepsy 3 (6)
First unprovoked seizure 10 (20)
figures of the follow-up MR imaging showing complete resolution,
Unspecified 7 (14) although this was not an essential reporting criterion in our quality
Seizure Classification assessment.
NCSE 17 (35)
CSE (focal onset) 13 (27)
3.3. MRI abnormalities
CSE (generalised) 1 (2)
CSE (unknown onset) 3 (6)
FMSE 2 (4) 3.3.1. MRI sequence
Focal SE (not further specified) 8 (16) Table 2 summarises SRMA according to the MRI sequences. The
Myoclonic SE 1 (2) sequence on which abnormality was most commonly reported was T2-
Unspecified 4 (8)
ASMs preceding onset of status epilepticus
weighted imaging in 42 patients (86%). Forty-one subjects (84%)
Phenytoin 2 (4) demonstrated T2 hyperintensity whereas hypointensity with ‘dark white
Phenobarbital 1 (2) matter’ was reported in one patient (4%). Although in 16 (33%) cases
Lamotrigine, phenytoin & levetiracetam 1 (2) field strength of MRI was not reported, all other cases specified that MRI
Not reported 31 (63)
was performed at 1.5T.
Treatment to terminate status epilepticus
PHT 4 (8) Time interval between the seizure and the MRI: Eight patients (16%) had
DZP 1 (2) an MRI conducted less than 24 h postictally, ten (20%) had an MRI
CBZ, CPZ & VPA 1 (2) between 1 and 2 days after the index seizure, and the MRI of five patients
MDZ, PPF, PHT, LEV, LCM, barbiturate coma 1 (2) (10%) was conducted between 2 and 3 days after the seizure. Two pa­
MDZ, PPF, PHT, LEV, LCM, VPA, CBZ, TPM, barbiturate 2 (4)
coma
tients (4%) had an MRI between 4 and 7 days post-ictus, and 4 patients
MDZ, PPF, PHT, LEV, TPM, CLB 1 (2) (8%) had an MRI performed after seven days or more. Ten patients
MDZ, PPF, PHT, LEV, VPA, TPM, barbiturate coma 1 (2) (20%) had an MRI during the ictus, whilst the initial MRI timing was not
MDZ, PPF, VPA, CBZ, PHB, barbiturate coma 1 (2) specified in 10 patients (20%). The median time interval to initial MRI
MDZ, VPA, LEV, PHT 1 (2)
was 24 h post-ictal (range: during ictus – 264 h after ictus).
MDZ, VPA, OXC 1 (2)
PHT & CBZ 1 (2) Between the index seizure and follow-up brain MRI showing com­
PHT & GBP 1 (2) plete resolution, there was a median time interval of 96.5 days (range:
PHT, CPZ & VPA 1 (2) ictal to 1800 days). Additionally, abnormal serial MRI scans were re­
PHT, LEV, GBP, TPM, CPZ 1 (2) ported in 12 patients (24%) ranging from 6 days to 70 days after the
PHT, PHB, CBZ, TPM, LEV, PPF 1 (2)
PHT, PHB, CBZ, TPM, LEV, thiopental 1 (2)
index seizure.
PHT, PHB, CBZ, TPM, LZP, barbiturate coma 1 (2) Anatomical distribution of SRMA: A summary of the anatomical dis­
PHT, PPF, CBZ, OXC, LZP, thiopental 1 (2) tribution of SRMA is presented in Table 3. Twenty-seven subjects (55%)
Pharmacologically induced coma 1 (2) exhibited unilateral SRMA whilst the remaining 22 (45%) had bilateral
Not reported 26 (53)
SRMA. There was a widespread of lobar location with cases reporting
Treatment Response
Responsive 10 (20) SRMA in every lobe. With respect to anatomical layers, the cortex and
Refractory 8 (16) hippocampus were the most commonly involved sites of SRMA.
Super-refractory 4 (8) A total of nine patients also had pre-existing structural abnormal­
Unspecified 27 (55) ities, such as tumours or vascular pathology, with six specifying a lobar
ASM = antiseizure medication; NCSE = non-convulsive status epilepticus; location. In four patients, the lobe of structural abnormality was at least
CSE = convulsive status epilepticus partially involved in the SRMA lesion, and one of these patients, who
FMSE = focal motor status epilepticus; PHT = phenytoin; DZP = diazepam; had a right parietal cavernoma, exhibited SRMA exclusively in the same
CBZ = carbamazepine; CPZ = clonazepam; VPA = valproate; GBP = gabapentin; lobe. Interestingly, all six subjects demonstrated SRMA ipsilateral to the
PHB = phenobarbital; TPM = topiramate; LEV = levetiracetam; original structural pathology.
LZP = lorazepam; PPF = Propofol; OXC = oxcarbazepine; MDZ = midazolam;
SRMA according to clinical and EEG characteristics: Overall, no specific
LCM = lacosamide; CLV = clobazam
patterns on the lateralisation or localisation of SRMA in relation to
subtypes of SE were evident. Nine (82%) patients presenting with focal

168
F.P. Mariajoseph et al.
Fig. 1. PRISMA Flowchart of article search, (n = number).
SE demonstrated unilateral SRMA (7 were ipsilateral to seizure focus; 2
were unspecified), although there was not enough data to identify trends
in patients with generalised SE (Table 4).
In total, 11 patients demonstrated specific unilateral interictal EEG
findings, and of these, 10 exhibited SRMA ipsilaterally (and one
exhibited bilateral SRMA), and 8 demonstrated SRMA involving the lobe
and location of EEG abnormality. In addition, two patients demon­
strated generalized periodic epileptiform discharges (GPEDs) on inter­
ictal EEG, and both these patients exhibited bilateral SRMA. Unilateral
ictal EEG findings specifying location were reported in 14 subjects, and
of these 11 displayed unilateral abnormalities ipsilateral to the ictal EEG
finding, and 3 patients had bilateral SRMA.
SRMA characteristics according to the duration of status epilepticus
are summarised in Supplementary Table 4. There were no obvious pat­
terns observed.
Response to therapy and SRMA: We also analysed SRMA in relation to
treatment response. Status epilepticus resolved with benzodiazepines or
standard ASM was defined as ‘responsive SE’ whilst patients who failed
to respond to ASM but responded to anaesthetic agents were classified as
‘refractory SE’. ‘Super-refractory SE’ was defined as the failure of an­
aesthetics to terminate SE. We found 10 responsive, 8 refractory, and 4
super-refractory SE patients whereas 27 cases could not be classified due
to insufficient data. Among ‘responsive SE’ patients, the cortical pattern
169
Seizure: European Journal of Epilepsy 92 (2021) 166–173
was the most frequent (6) followed by subcortical (2) and claustrum
patterns (2). Cortical atrophy was evident in 6 patients as long-term
sequelae. Hippocampal involvement was the most frequent stereotypic
pattern seen in refractory SE (in 5) and four subjects went on to develop
bilateral hippocampal atrophy on follow-up. Bilateral claustrum
involvement was seen in three subjects whilst only one had signal
changes in the splenium. Interestingly, all four patients in the super
refractory SE group demonstrated bilateral claustrum involvement.
Hyperintensities in T2, FLAIR, and DWI sequences were seen across all
three groups.
SRMA according to aetiology: It is noteworthy that nine patients in the
pooled data were diagnosed with new-onset refractory status epilepticus
(NORSE) and eight of them had SRMA involving the claustrum bilater­
ally whilst the remaining patient demonstrated signal changes in the
splenium. Outside the NORSE group, only one patient diagnosed as
possible febrile infection-related epilepsy syndrome (FIRES) had bilat­
eral claustrum involvement (see Supplementary Table 1). Though this
data has limitations, these results suggest a possible association between
NORSE and SRMA in the claustrum. Alternatively, it could be a reflec­
tion of the severity of SE as all four subjects with super-refractory SE
demonstrated signal changes in the claustrum bilaterally.
Long term sequelae: Cortical atrophy alone was present in six subjects
(12%), whilst hippocampal sclerosis was reported in 9 patients (18%).
F.P. Mariajoseph et al. Seizure: European Journal of Epilepsy 92 (2021) 166–173

Table 2 Table 3
Summary of MRI findings according to sequence. Summary of anatomical lobar and layer location of SRMA.
MRI features according to sequence Number of cases (%) Lobar location of MRI abnormalities Number of cases (%)

DWI Unilateral 25 (51)

Unknown/Not Done 13 (27) Frontal 2 (4)
No Change/Not Restricted 4 (8) Frontoparietal 1 (2)
Restricted diffusion 32 (65) Frontotemporal & Parietooccipital 1 (2)
ADC Hippocampus 4 (8)
Unknown/Not Done 16 (33) Hippocampus & Pulvinar 3 (6)
No Change 9 (18) Occipital, Parahippocampus & Uncus 1 (2)
Decreased 21 (43) Parietal 3 (6)
Increased 3 (6) Parietooccipital 1 (2)
FLAIR Parietooccipital & Pulvinar 1 (2)
Unknown/Not Done 19 (39) Parietooccipital, Pulvinar & Hippocampus 1 (2)
Hyperintense 29 (59) Temporal, Hippocampus & Uncus 1 (2)
Hyperintense + Dark WM 1 (2) Temporooccipital 2 (4)
T2 Temporoparietal 3 (6)
Unknown/Not Done 6 (12) Unspecified 1 (2)
Normal 1 (2) Bilateral 22 (45)
Hyperintense 41 (84) Cingulate Cortex & Unilateral Frontotemporal 1 (2)
Hyperintense + Dark WM 1 (2) Cingulate Gyrus 1 (2)
T1 Claustrum 4 (8)
Unknown/Not Done 30 (61) Claustrum & External capsule 2 (4)
Normal 8 (16) Claustrum & Insular Cortex 1 (2)
Hypointense 11 (22) Claustrum & Unilateral Thalamus 1 (2)
PD Cuneus & Precuneus 1 (2)
Unknown/Not Done 48 (98) Diffuse 1 (2)
Hyperintense 1 (2) Frontotemporal & Hippocampus 1 (2)
Post Contrast Hemisphere Cortex 1 (2)
Unknown/Not Done 25 (51) Hippocampus 4 (8)
No Change/Enhancement 15 (31) Temporal & Unilateral Parietal 1 (2)
Enhancement 9 (18) Splenium, Hippocampus, Claustrum, Insula & Basal 1 (2)
Ganglia 2 (4)
DWI = Diffusion weighted imaging; ADC = Apparent diffusion coefficient; Unspecified 2 (4)
FLAIR = Fluid-attenuated inversion recovery; PD = Post diffusion Central 2 (4)
Splenium
One patient (2%) developed both hippocampal sclerosis and cortical Layer location of MRI abnormalities Number of cases (%)
atrophy. Two subjects (4%) demonstrated cortical laminar necrosis on Cortical 8 (16)
follow-up MRI. Complete resolution of SRMA without any residual Cortical + Subcortical 7 (14)
findings was observed in eight patients (16%), although long-term Cortical + Subcortical + Hippocampus 2 (4)
sequelae were not reported in 23 subjects (47%). These findings were Cortical + Thalamus 2 (4)
Cortical + Thalamus + Hippocampus 1 (2)
observed at the time of complete resolution of SRMA (median 96.5 days;
Claustrum 6 (12)
range ictal to 1800 days). Claustrum + External Capsule 2 (4)
Composite patterns: Through a comprehensive review of all available Hippocampus 8 (16)
MRI images, we identified stereotypical presentations of SRMA, which Hippocampus + Thalamus 3 (6)
we have labelled “composite patterns”. A total of 5 patterns were Splenium 2 (4)
Subcortical 2 (4)
identified: (1) predominant cortical (with or without subcortical, lep­ Unspecified 6 (12)
tomeningeal or thalamic involvement). (2) Hippocampal (with or
without cortical, subcortical, leptomeningeal or thalamic involvement),
(3) claustrum, (4) predominant subcortical, and (5) splenium involve­ early as during the ictus with the median length of time from appearance
ment (Supplementary table 3 & Fig. 2). to complete resolution of SRMA being 96.5 days. Furthermore, there was
Predominant cortical (gyral) signal changes (with or without also an apparent association between EEG findings and lateralisation of
subcortical, leptomeningeal or thalamic involvement) were the most SRMA in the contexts of both ictal and interictal EEG. Additionally,
salient pattern that we observed (17 patients). Eleven of these patients cortical atrophy and hippocampal sclerosis were the most common long-
demonstrated this pattern unilaterally, whilst the remaining six term sequelae evident on the MRI after reversal of seizure-related MRI
demonstrated a bilateral pattern. The second most frequently identified changes.
composite pattern was hippocampal SRMA (with or without cortical,
subcortical or thalamic involvement) observed in 10 patients. Further­ 4.1. Clinical relevance
more, bilateral signal changes in the claustrum were noted in 9 patients
whereas SRMA involving the splenium and a predominant subcortical We determined composite patterns through an extensive analysis of
pattern were identified in two subjects each. all available MRI data, including provided images. MRI abnormalities
predominantly cortical in nature were most commonly observed (17
4. Discussion patients), followed by hippocampal SRMA (14 patients) and the claus­
trum bilaterally (9 patients). Abnormalities were most frequently seen as
Through our analysis of a total of 35 subjects who suffered from SE, hyperintensity on a T2-weighted sequence, although there was consid­
we determined five stereotypical SRMA patterns involving the (1) cor­ erable paucity in the reporting of MRI data in some sequences. Post-
tex, (2) subcortical region, (3) hippocampus, (4) claustrum, and (5) contrast and FLAIR imaging, for example, were reported in less than
splenium. Of these, MRI abnormalities most commonly manifested in half of the cases. Additionally, T1-weighted imaging data was only
the cortex, with a widespread of lobar locations. Abnormalities were available in 19 patients, with 11 (22%) demonstrating signal change and
most frequently reported on T2-weighted sequences followed by DWI/ eight (16%) showing no abnormality. As such, all MRI sequences should
ADC and FLAIR images. We also observed that SRMA can manifest as be studied in investigating status epilepticus in order to get a complete

170
F.P. Mariajoseph et al. Seizure: European Journal of Epilepsy 92 (2021) 166–173

Table 4 the vast majority reporting one MRI scan several months following the
Summary of SRMA according to seizure classification. seizure. As a result, we were only able to report on when the normal MRI
MRI abnormalities according to seizure classification Number of cases (%) scan was performed, rather than the exact timeline for normalisation of
SRMA. Consequently, the timeframes that we report are by no means
NCSE
Unilateral 14 (29) conclusive, although we believe they have provided valuable estimates
Frontotemporal & Parietooccipital 1 (2) for clinicians about the duration of SRMA and its eventual reversal.
Hippocampus 2 (4) Nonetheless, we would suggest that other pathologies be considered
Hippocampus & Pulvinar 3 (6) before confirming SRMA, especially if the lesion persists for extended
Occipital & Parahippocampus & Uncus 1 (2)
Parietal 1 (2)
periods without any evidence of regression.
Parietooccipital & Pulvinar 1 (2) When compared with a recent systematic review of SRMA following
Parietooccipital & Hippocampus & Pulvinar 1 (2) single seizures [28], there are several differences between the charac­
Temporoparietal 3 (6) teristics of SRMA in single seizures and status epilepticus. Following a
Temporal & Hippocampus & Uncus 1 (2)
single seizure, abnormalities have been reported to resolve within days
Bilateral 3 (6)
Cingulate gyri 1 (2) to weeks of onset, whilst we observed that SRMA takes longer to reverse
Cingulate cortex & Unilateral frontotemporal 1 (2) in SE and is more likely to take weeks to months before complete nor­
Temporal & Unilateral Parietal 1 (2) malisation. Status epilepticus patients also had numerous reports of
CSE (focal onset) long-term sequelae, after resolution of SRMA, such as hippocampal
Unilateral (ipsilateral) 5 (10)
Frontoparietal 1 (2)
sclerosis, cortical laminar necrosis, and cortical atrophy. This was not
Hippocampus 2 (4) characteristic in SRMA following single seizures. Additionally, in single
Parietooccipital 1 (2) seizure patients, the subcortical region was reported to be the most
Temporoparietal 1 (2) frequent location of SRMA, whereas, in the context of SE subcortical
Bilateral 7 (14)
lesions were far less common. Nonetheless, there are some similarities as
Frontotemporal & Hippocampus 1 (2)
Hippocampus 4 (8) well. In both single seizures as well as status epilepticus, SRMA was most
Diffuse 1 (2) commonly reported in T2-weighted imaging. Furthermore, the hippo­
Splenium, Hippocampus, Claustrum, Insula & Basal 1 (2) campus was not an infrequent location of manifestation in both contexts,
Ganglia although, single seizures largely demonstrated unilateral abnormalities
CSE (generalised onset)
whilst hippocampal involvement in SE patients was predominantly
Central 1 (2)
Splenium 1 (2) bilateral.
CSE (unknown onset) There was very limited information on seizure characteristics and
Unilateral 1 (2) therefore we were restricted in our exploration of their association with
Frontal 1 (2)
SRMA location. In focal SE, we noted that SRMA were unilateral as well
Bilateral 3 (6)
Hemisphere Cortex 1 (2) as bilateral in nature. Furthermore, in the current study, we were unable
Claustrum & External Capsule 1 (2) to examine the laterality of SRMA in generalised SE due to a paucity of
Cuneus & Precuneus 1 (2) data. Conversely, in single seizure patients, focal seizures are more likely
FMSE to present as unilateral SRMA and patients with generalised seizures are
Unilateral (ipsilateral) 2 (4)
more likely to exhibit bilateral signal changes [28].
Frontal 1 (2)
Temporooccipital 1 (2) The main clinical dilemma posed by SRMA is the differentiation from
Focal SE unspecified underlying cerebral pathologies responsible for SE. There are no
Unilateral (laterality unknown) 2 (4) guidelines on how to approach this question in a systematic manner. It is
Parietal 1 (2)
also unclear how often and how soon the imaging should be repeated.
Unspecified 1 (2)
Bilateral 6 (12)
Though we are unable to provide definitive guidelines due to many
Claustrum 4 (8) limitations of the literature, the stereotypic MRI patterns delineated by
Claustrum & Insular Cortex 1 (2) us may help clinicians recognize this phenomenon early. Further
Claustrum & Unilateral Thalamus 1 (2) research is needed to establish clear guidelines and management
Unspecified SE
pathways.
Unilateral 1 (2)
Parietal 1 (2)
Bilateral 2 (4) 4.2. Pathophysiology
Unspecified 2 (4)
Central 1 (2)
It has been postulated that increased blood flow, metabolic demand
Splenium 1 (2)
Myoclonic SE and neurotransmitter release can be attributed to the transient abnor­
Bilateral 1 (2) malities seen in patients exhibiting SRMA, especially when abnormal­
Claustrum & External Capsule 1 (2) ities coincide with the location of the seizure focus [16,29]. Epileptic
NCSE = Non-convulsive status epilepticus; CSE = Convulsive status epilepticus; seizure activity causes increased neurotransmitter release as well as
FMSE = Focal motor status epilepticus; SE = status epilepticus raised metabolic requirements, and blood flow to the seizure focus is
markedly increased in an attempt to meet these energy demands. This is
picture. Likewise, serial MRI scans are also recommended as they will ultimately insufficient and leads to dependence on anaerobic respira­
provide more accurate information on the time course of reversible tion, which results in the accumulation of lactic acid, carbon dioxide and
signal changes. On another note, we did not observe any association free radicals [30], causing a release of apoptotic enzymes and a break­
between the duration of the seizure and SRMA characteristics. None­ down of the neuronal cell membrane[2].
theless, we found that a pattern did emerge between SRMA and EEG As aforementioned, the hippocampus is commonly involved in SRMA
findings. In both the instances of interictal and ictal EEG, patients with which may be explained by an increased number of glutamate receptors
focal EEG findings were likely to exhibit SRMA ipsilaterally. Addition­ in the hippocampus, rendering it more susceptible to cytotoxic damage
ally, we observed a possible association between signal alterations [31,32] Additionally, it has been suggested that, especially in the
involving the claustrum bilaterally and NORSE. context of prolonged seizures, the hippocampus can be vulnerable to
In this study, serial imaging was available in a minority of cases, with atrophic changes [33]. Ultimately these conclusions were supported by
our findings, as we found 14 patients (28.5%) with hippocampal SRMA,

171
F.P. Mariajoseph et al. Seizure: European Journal of Epilepsy 92 (2021) 166–173

Fig. 2. Composite patterns of MRI signal abnormalities seen in status epilepticus. Cortical pattern- A: focal diffusion restriction involving cortical gyri, B: ipsilateral
pulvinar signal change on fluid-attenuated inversion recovery imaging; 2. Subcortical pattern-C: hyperintensity in the subcortical left insular region on fluid-
attenuated inversion recovery imaging; 3. Splenial pattern-D: diffusion restriction of the splenium, E: corresponding decreased signal intensity on the apparent
diffusion coefficient image; 4. Hippocampal pattern-F: diffusion restriction in hippocampi bilaterally, G: corresponding decreased signal intensity on the apparent
diffusion coefficient image, H: hyperintensity of hippocampi bilaterally on fluid-attenuated inversion recovery imaging, I: contrast enhancement of hippocampi
bilaterally on T1-weighted imaging (post-contrast); J: gliosis and encephalomalacia in the right fontal region due to previous head trauma. This lesion is the aetiology
of status epilepticus. Note bilateral signal changes in hippocampi due to status epilepticus away from the lesion.

as well as 10 patients (20%) who went on to develop hippocampal
sclerosis.
With regards to MRI sequences specifically, diffusion restriction on
DWI/ADC signifies cytotoxic oedema [4], whilst T2-weighted imaging
and FLAIR abnormalities reflect vasogenic oedema.[2] During epileptic
seizures, changes observed in ADC/DWI occur simultaneously to signal
changes in T2 and FLAIR, which suggest that the pathophysiological
mechanism of these abnormalities cannot be explained by ischaemic
causes or hypoxia in which these changes usually appear sequentially.
4.3. Strengths
There are a number of strengths inherent to this study. This review
was conducted in strict accordance with the PRISMA guidelines. Eligi­
bility criteria for selection into this review were stringently imple­
mented. Complete resolution of SRMA was ensured, as well as the
exclusion of provoked seizures, cluster seizures and single seizures,
minimising the impact of potential confounders in our study. Addi­
tionally, there is a widespread of patient characteristics represented
within this study, rendering our findings applicable in different settings.
Demographically, patients originated from eleven countries with the
youngest and oldest subjects being aged 2 years and 79 years
respectively.
resolution of SRMA, as the vast majority of cases only reported the time
point at which SRMA disappeared from MRI. An additional limitation
inherent to reviewing status epilepticus is the changes in definitions that
have taken place over the years. It is possible that some relevant cases
were not included in our study because they were not considered SE at
the time of publication, however, we attempted to minimise this by
carefully considering the length of seizure episodes and applying current
definitions to them, although some cases may still have been missed due
to inadequate information provided in the publication.
Additionally, there was considerable heterogeneity in the reporting
of data. Although 82% of cases provided images of the SRMA itself, only
57% presented figures of the resolved MRI scan, which may indeed have
provided useful information. Not all MRI sequences were reported.
Standard MRI sequences were not mentioned in several studies, with
almost two-thirds of cases failing to report T1 weighted imaging results
and more than half excluding FLAIR sequence. Similarly, long-term
clinical outcomes of patients were infrequently reported, as was there
a paucity in the reporting of management, aetiology and comorbidities.
Further, there was variation in SE classifications, as well as very limited
information regarding seizure focus, which ultimately hindered our
analysis of their associations with SRMA characteristics.
4.5. Gaps in knowledge

There is still considerable uncertainty surrounding SRMA, and many

4.4. Limitations
knowledge gaps remain. The incidence of SRMA itself, which would
indeed be clinically relevant, is not known. Similarly, although there are
The shortcomings of this study are primarily attributed to the lack of
some theories postulated, there is currently no conclusive explanation
standardisation in clinical and imaging data as well as the small sample
for the pathophysiological mechanisms of SRMA. Additionally, due to
size. As aforementioned, it was necessary to implement strict eligibility
insufficient data in this study, we were unable to draw conclusions about
criteria to reduce confounding factors as well as ensure uniformity in the
the associations between SRMA and seizure characteristics such as
outcome. An undesirable consequence of this rigid approach was the
seizure focus. It is also unknown whether SRMA holds any significance
exclusion of a large proportion of cases. From the 95 cases reported
in patient outcomes and prognosis. Furthermore, any predictive factors,
within the 19 publications that were selected, only 49 were eligible for
such as demographics or clinical characteristics, that make the devel­
analysis, although the stringent implementation of the eligibility criteria
opment of SRMA more likely, are not known.
was deemed necessary to ensure the reliability of our results.
Although serial scanning was performed in 12 patients (20%), this
number was insufficient to make definitive estimations on the time to

172
F.P. Mariajoseph et al.
4.6. Future directions
Further investigation on SRMA is warranted given its clinical im­
plications. Underrecognition of this entity may lead to unnecessary in­
vestigations such as brain biopsy. Continued research into the
pathophysiology of this phenomenon would be useful to gain a better
clinical understanding. Studies with larger sample sizes would also
provide additional details and insight into the characteristics of SRMA in
a variety of settings and presentations. Additionally, serial scanning at
pre-determined time points would facilitate a more accurate estimate of
the time frames for both the initial occurrence and subsequent resolu­
tion of SRMA. Finally, studies comparing cohorts of patients who
develop SRMA with cohorts who do not, may provide information on
predisposing factors associated with the appearance of SRMA.
5. Conclusion
In the published literature on status epilepticus, SRMA manifested
heterogeneously with frequent involvement on T2-weighted sequences.
Abnormalities were noted to appear as early as during the ictus and as
late as 11 days post-ictal, and the median time to resolution was 96.5
days. The cortex, hippocampus, and claustrum were common locations
of SRMA, and long-term sequelae, such as hippocampal sclerosis,
cortical atrophy and cortical laminar necrosis were also noted. Interictal
and ictal EEG findings were also useful in predicting the lateralisation of
abnormalities. However, due to considerable limitations of the litera­
ture, no definitive conclusions can be drawn to propose a clear man­
agement pathway for those patients presenting with SRMA. This study
highlights the need for more research into SRMA, specifically relating to
serial scanning and prognostic implications, in order to formulate
practical guidelines to rationally manage this clinical dilemma.
Funding
This research did not receive any specific grant from funding
agencies in the public, commercial, or not-for-profit sectors.
Declaration of Competing Interest
We know of no conflicts of interest associated with this manuscript,
and there has been no significant financial support for this work that
could have influenced its outcome.
Supplementary materials
Supplementary material associated with this article can be found, in
the online version, at doi:10.1016/j.seizure.2021.09.002.
References
[1] Sánchez S, Rincon F. Status epilepticus: epidemiology and public health needs.
J Clin Med 2016;5(8):71.
[2] Briellmann RS, Wellard RM, Jackson GD. Seizure-associated abnormalities in
epilepsy: evidence from MR imaging. Epilepsia 2005;46(5):760–6.
[3] Cole AJ. Status epilepticus and periictal imaging. Epilepsia 2004;45:72–7.
[4] Kim SE, Lee BI, Shin KJ, Ha SY, Park J, Park KM, et al. Characteristics of seizure-
induced signal changes on MRI in patients with first seizures. Seizure 2017;48:
62–8.
[5] Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, et al. The
PRISMA statement for reporting systematic reviews and meta-analyses of studies
that evaluate health care interventions: explanation and elaboration. J Clin
Epidemiol 2009;62(10):e1–34.
173
Seizure: European Journal of Epilepsy 92 (2021) 166–173
[6] Trinka E, Cock H, Hesdorffer D, Rossetti AO, Scheffer IE, Shinnar S, et al.
A definition and classification of status epilepticus–report of the ILAE task force on
classification of status epilepticus. Epilepsia 2015;56(10):1515–23.
[7] Fisher RS, Cross JH, French JA, Higurashi N, Hirsch E, Jansen FE, et al. Operational
classification of seizure types by the international league against epilepsy: position
Paper of the ILAE commission for classification and terminology. Epilepsia 2017;58
(4):522–30.
[8] Murad MH, Sultan S, Haffar S, Bazerbachi F. Methodological quality and synthesis
of case series and case reports. BMJ Evid Based Med 2018;23(2):60–3.
[9] Bauer G, Gotwald T, Dobesberger J, Embacher N, Bauer R, Seppi K, et al. Transient
and permanent magnetic resonance imaging abnormalities after complex partial
status epilepticus. Epilepsy Behav 2006;8(3):666–71.
[10] Canas N, Soares P, Calado S, Pestana R, Ribeiro C, Vale J. Pathophysiology and
long-term outcome of reversible tumor-like lesions induced by presenting status
epilepticus. J Neuroimaging 2010;20(2):169–74.
[11] Chhetri SK, Mathur S, Soh C, Gosal D. Seizure-induced MRI changes mimicking
metastatic brain disease. J Clin Neurosci 2014;21(5):869–72.
[12] Cianfoni A, Caulo M, Cerase A, Della Marca G, Falcone C, Di Lella GM, et al.
Seizure-induced brain lesions: a wide spectrum of variably reversible MRI
abnormalities. Eur J Radiol 2013;82(11):1964–72.
[13] Huang YC, Weng HH, Tsai YT, Huang YC, Hsiao MC, Wu CY, et al. Periictal
magnetic resonance imaging in status epilepticus. Epilepsy Res 2009;86(1):72–81.
[14] Ishiguro T, Nitta M, Komori T, Maruyama T, Muragaki Y, Kawamata T. Transient
focal magnetic resonance imaging abnormalities after status epilepticus showed
11C-methionine uptake with positron emission tomography in a patient with
cerebral cavernous malformation. World Neurosurg 2018;114:43–6.
[15] Juhász C, Scheidl E, Szirmai I. Reversible focal MRI abnormalities due to status
epilepticus. An EEG, single photon emission computed tomography, transcranial
Doppler follow-up study. Electroencephalogr Clin Neurophysiol 1998;107(6):
402–7.
[16] Kim JA, Chung JI, Yoon PH, Kim DI, Chung TS, Kim EJ, et al. Transient MR signal
changes in patients with generalized tonicoclonic seizure or status epilepticus:
periictal diffusion-weighted imaging. Am J Neuroradiol 2001;22(6):1149–60.
[17] Kramer R, Lüders H, Lesser R, Weinstein M, Dinner D, Morris H, et al. Transient
focal abnormalities of neuroimaging studies during focal status epilepticus.
Epilepsia 1987;28(5):528–32.
[18] Lansberg MG, O’Brien M, Norbash A, Moseley M, Morrell M, Albers G. MRI
abnormalities associated with partial status epilepticus. Neurology 1999;52(5):
1021.
[19] Milligan TA, Zamani A, Bromfield E. Frequency and patterns of MRI abnormalities
due to status epilepticus. Seizure 2009;18(2):104–8.
[20] Senn P, Lövblad KO, Zutter D, Bassetti C, Zeller O, Donati F, et al. Changes on
diffusion-weighted MRI with focal motor status epilepticus: case report.
Neuroradiology 2003;45(4):246–9.
[21] Szabo K, Poepel A, Pohlmann-Eden B, Hirsch J, Back T, Sedlaczek O, et al.
Diffusion-weighted and perfusion MRI demonstrates parenchymal changes in
complex partial status epilepticus. Brain 2005;128(6):1369–76.
[22] Chevret L, Husson B, Nguefack S, Nehlig A, Bouilleret V. Prolonged refractory
status epilepticus with early and persistent restricted hippocampal signal MRI
abnormality. J Neurol 2008;255(1):112–6.
[23] Eguchi K, Tsuzaka K, Yabe I, Sasaki H. Reversible splenial lesion in new-onset
refractory status epilepticus: a case report. Clin Neurol Neurosurg 2019;183:
105392.
[24] Meletti S, Slonkova J, Mareckova I, Monti G, Specchio N, Hon P, et al. Claustrum
damage and refractory status epilepticus following febrile illness. Neurology 2015;
85(14):1224–32.
[25] Mizutani AU, Shindo A, Arikawa S, Shimada T, Matsuura K, Ikejiri K, et al.
Reversible splenial lesion in a patient with new-onset refractory status epilepticus
(NORSE). Eneurologicalsci 2020;18:100220.
[26] Mumoli L, Labate A, Palamara G, Sturniolo M, Gambardella A. Reversible
symmetrical external capsule hyperintensity as an early finding of autoimmune
encephalitis. Neurol Sci 2014;35(7):1147–9.
[27] Sperner J, Sander B, Lau S, Krude H, Scheffner D. Severe transitory encephalopathy
with reversible lesions of the claustrum. Pediatr Radiol 1996;26(11):769–71.
[28] Mariajoseph FP, Muthusamy S, Amukotuwa S, Seneviratne U. Seizure-induced
reversible MRI abnormalities in patients with single seizures: a systematic review.
Epileptic Disord 2021;23(4):552–62.
[29] Sutula TP. Experimental models of temporal lobe epilepsy: new insights from the
study of kindling and synaptic reorganization. Epilepsia 1990;31:S45–54.
[30] Dingledine R, Varvel NH, Dudek FE. When and how do seizures kill neurons, and is
cell death relevant to epileptogenesis? Adv Exp Med Biol 2014;813:109–22.
[31] Khaspekov L, Stastný F, Viktorov I, Lisý V. Cytotoxic effect of glutamate and its
agonists on mouse hippocampal neurons. J Brain Res 1990;31(5):635–43.
[32] Margerison J, Corsellis J. Epilepsy and the temporal lobes. Brain 1966;89(3):
499–530.
[33] Pohlmann-Eden B, Gass A, Peters C, Wennberg R, Bluemcke I. Evolution of MRI
changes and development of bilateral hippocampal sclerosis during long lasting
generalised status epilepticus. J Neurol Neurosurg Psychiatry 2004;75(6):898–900.