3.0 T MRI of 2000 consecutive patients with localisation-related epilepsy

Abstract

Objectives

Clinical guidelines suggest that all patients diagnosed with localised seizures should be investigated with MRI to identify any epileptogenic structural lesions, as these patients may benefit from surgical resection. There is growing impetus to use higher field strength scanners to image such patients, as some evidence suggests that they improve detection rates. We set out to review the detection rate of radiological abnormalities found by imaging patients with localised seizures using a high-resolution 3.0 T epilepsy protocol.

Methods

Data were reviewed from 2000 consecutive adult patients with localisation-related epilepsy referred between January 2005 and February 2011, and imaged at 3.0 T using a standard epilepsy protocol.

Results

An abnormality likely to be related to seizure activity was identified in 403/2000 (20.2%) patients, with mesial temporal sclerosis diagnosed in 211 patients. 313/2000 (15.6%) had lesions potentially amenable to surgery. Abnormalities thought unrelated to seizure activity were found in 324/2000 (16.1%), with 8.9% having evidence of ischaemic disease.

Conclusions

Since the introduction of the then National Institute for Clinical Excellence guidelines in 2004, the detection rate of significant pathology using a dedicated 3.0 T epilepsy protocol has not fallen, despite the increased numbers of patients being imaged. This is the largest study of epilepsy imaging at 3.0 T to date and highlights the detection rates of significant pathology in a clinical setting using a high-strength magnet. The prevalence of ischaemic disease in this population is significantly higher than first thought, and may not be incidental, as is often reported.

In developed countries the incidence of epilepsy is around 50 per 100 000 people per year, with a prevalence of between 0.5% and 1% [1]. Neuroimaging has a valuable role in these patients to identify epileptogenic structural brain lesions that may be amenable to surgery. Indeed, there are now guidelines published by the National Institute for Health and Clinical Excellence (NICE) [2] and the International League Against Epilepsy [3,4] suggesting that all patients with localised seizures should undergo MRI to assess for structural abnormalities.

The epidemiology of epileptogenic structural lesions is likely to be different depending on the age at presentation, and the rate of detecting these abnormalities will depend on the selection criteria for imaging and the type of MRI performed. In paediatric studies, the rate of seizure-related abnormalities ranges between 13% and 51% [5-10]. In adults the rates are similar (13–55%), but the number of patients studied is relatively low and the selection criteria for MRI vary substantially [11-13]. The detection rate is higher in patients with refractory epilepsy; in two of the larger studies structural lesions were found in 82% and 86% of patients [14,15].

The rate of abnormal findings may also be dependent on the strength of the magnet used. The increased signal to noise ratio and superior anatomical detail that 3.0 T MRI provides has been shown to improve sensitivity in detecting mesial temporal sclerosis, vascular malformations and, most noticeably, cortical malformations [16-19]. Recent evidence suggests that the detailed anatomy of the hippocampus on 3.0 T MRI correlates well with surgical pathological findings and is superior to 1.5 T [20].

In 2005 we began imaging all patients referred with focal seizures with 3.0 T MRI using a standard epilepsy protocol. Here we present the imaging findings from 2000 consecutive patients between 2005 and 2011.

Methods and materials

Our unit is a regional clinical neuroscience facility receiving referrals from a catchment population of 2.3 million. This retrospective study was registered with the audit department and included 2000 consecutive patients referred by neurologists with localised seizures between January 2005 and February 2011. The protocol is only used for patients with a diagnosis of localised epilepsy. Patients with generalised epilepsy syndromes and patients presenting with first seizures were imaged by another means and are not included in this study.

All patients were imaged using a high-resolution epilepsy protocol (Table 1) on a superconducting 3.0 T system (Intera 3.0T; Philips Medical Systems, Best, Netherlands) at the University of Sheffield MRI facility. The coronal imaging was performed angled specifically for the temporal lobes. Intravenous contrast was only used if a mass lesion was detected.

Table 1. The sequences of our 3.0 T epilepsy protocol.

Sequence	Plain	Slice thickness (mm)	Time (min:s)
FSE T2	Axial	5	4:47
FSE T2	Coronal (TL)	2	4:18
T2 weighted FLAIR	Coronal (TL)	2	5:47
GE T1 volume	Coronal (TL)	0.8	12:47
IR T1	Coronal (TL)	2	4:47
T2 relaxometry	Coronal (TL)	5	4:47
DWI sense	Coronal (TL)	3	0:48

DWI, diffusion-weighted imaging; FLAIR, fluid-attenuated inversion recovery; FSE, fast spin echo; GE, gradient echo; IR, inversion recovery; TL, non-standard temporal lobe angulation.

The MRI examinations were all reported by experienced consultant neuroradiologists who had access to the clinical information on the request card at the time of reporting (TH, DJAC, SCC, CAJR, RB, NH and PDG). Those patients with abnormal findings were discussed in the clinical multidisciplinary epilepsy meeting where a decision was made as to whether the pathology was related or unrelated to seizure activity based on the clinical history and electrophysiological results. If the lesion was thought to be amenable to surgery the patient was placed into a programme of monitoring and the subsequent decision on surgery was based on comorbidity, pharmacological seizure control and patient wishes.

Results

In total there were 1078 (54%) female and 922 (46%) male patients imaged. The interquartile age range was 23 years (range 25–48 years) and the full age range was 67 years (range 16–83 years). Figure 1 shows the number of examinations performed per year and the age ranges of those imaged.

Figure 1.

The proportion of examinations performed each year since the National Institute for Health and Clinical Excellence guidelines were introduced (*only includes the first 18 examinations in 2011) and the ages of the patients referred (in years).

The MRI examinations were reported as normal in 1219/2000 (61%) cases and abnormal in 727/2000 (36.4%). In 54/2000 (2.7%) the examination was non-diagnostic, because of either patient claustrophobia or movement artefact. For the abnormal examinations, the clinical team made a decision whether the findings were either related to seizure activity (403/2000; 20.2%) or unrelated to seizure activity (324/2000; 16.2%). Figure 2 shows the proportion of abnormal findings (both related and unrelated to seizures) with respect to the age of the patient.

Figure 2.

The proportion of related and unrelated abnormal findings correlated to the age of the patient. The percentages are of the total number of patients in that age range.

In the 403 patients with probable seizure-related abnormalities, mesial temporal sclerosis (MTS) (53.6%), cortical malformations (18.3%), vascular malformations (7.1%) and tumours (5.1%) were the main pathologies encountered (Table 2). A number of patients (10.9%) had evidence of encephalomalacia, gliosis or previous infarctions. There were 211 patients with isolated MTS, with 10 patients reported to have bilateral sclerosis. Another seven patients had MTS coexisting with another pathology (two vascular malformations, four cortical dysplasias and one tumour). There were another seven patients where two pathologies were found: two different cortical malformations in six patients and a nodular heterotopia and coexistent ganglioglioma in one patient.

Table 2. A summary of the 404 patients with abnormalities likely to be related to seizures.

Abnormality	Number	Percentage of abnormal	Percentage of total
MTS	211	53.6	10.6
Cortical malformations	72	18.3	3.6
Focal cortical dysplasia	36	9.1	1.8
Heterotopia	28	7.1	1.4
Polymicrogyria	4	1.0	0.2
Schizencephaly	4	1.0	0.2
Vascular malformations	28	7.1	1.4
Cavernoma	17	4.3	0.9
Arteriovenous malformation	10	2.5	0.5
Developmental venous anomaly	1	0.3	0.1
Tumours	20	5.1	1.0
Gangliogliomas	6	1.5	0.3
DNET	6	1.5	0.3
Astrocytomas	3	0.8	0.2
Oligodendrogliomas	1	0.3	0.1
Hypothalamic hamartoma	2	0.5	0.1
Meningioma	2	0.5	0.1
Unclassified low-grade tumour or cortical malformation	18	4.6	0.9
Phakomatoses	6	1.5	0.3
Tuberous sclerosis	4	1.0	0.2
Neurofibromatosis I	1	0.3	0.1
Sturge–Weber syndrome	1	0.3	0.1
Encephalomalacia or gliosis	38	9.6	1.9
Previous Infarction	5	1.3	0.3
Encephalitis	3	0.8	0.2
Tuberculomas	1	0.3	0.1
Encephalocoele	1	0.3	0.1

DNET, dysembryoplastic neuroepithelial tumour; MTS, mesial temporal sclerosis.

In this study there were 313/2000 (15%) patients with abnormalities that were considered potentially amenable to surgery. These included MTS, vascular malformations, focal cortical dysplasias and tumours (Figure 3). To date 51 patients have proceeded to surgery (average age 36.8 years; range 17–65 years). Table 3 shows the pathology detected in these patients. Two patients had surgery following normal MRI as electroencephalography and/or single-photon emission CT (SPECT) localised the seizures. Of the 49 patients found to have positive findings on MRI, 47 had histopathology reports available. The MRI findings agreed with the pathology results in 45/47 (96%) cases. In two patients histopathology confirmed focal cortical dysplasia with a previous working MRI diagnosis of dysembroplastic neuroepithelial tumour.

Figure 3.

The proportion of potentially resectable abnormalities by age. MTS, mesial temporal sclerosis.

Table 3. Summary of the 324 patients who had abnormalities thought unlikely to be related to seizures.

Abnormality	Number
Cerebrovascular disease
Infarct	32a
Ischaemia	136a
Demyelination
Definite	4
Probable	10
Non-specific white matter lesions	23
Atrophy
Supratentorial	33
Cerebellar	7
Vascular malformations
Cavernoma	9
Aneurysm	5
Tumours
Meningioma	9
Pituitary	5
L'Hermitte Duclos	1
Epidermoid	1
Cysts
Arachnoid	6
Colloid	2
Pineal	1
Porencephalic	1
Post-traumatic change	15
Posterior fossa abnormalities
Chiari Type I malformation	2
Cerebellar dysplasia	1
Cerebellopontine atrophy	1
Widened vestibular aqueduct	1
Kinked brain stem	1
Sinus disease	2
Previous surgery	2
Hydrocephalus	2
Neonatal vascular insult	13
Neonatal hypoglycaemia	1

^aIncludes those with ischaemia and an additional diagnosis.

In the 324 patients with abnormalities thought to be unrelated to seizure activity, findings included ischaemic changes, previous infarcts, cerebellar atrophy, unrelated tumours or vascular malformations and unexplained white matter hyperintensities (Table 4).

Table 4. The MRI diagnosis of the 51 patients who underwent surgery.

MRI diagnosis	Number
MTS	29
Low-grade tumours	8a
AVMs	5
Cortical dysplasias	5
Cavernoma	1
Tuberous sclerosis	1
Normal	2

AVMs, arteriovenous malformations; MTS, mesial temporal sclerosis.

^aTwo were found to have a cortical dysplasia on histology.

Discussion

This review of 2000 consecutive patients imaged with 3.0 T MRI for localised seizures is the largest study to date. It represents our current clinical practice since the introduction of the NICE guidelines in 2004. The patients included in this study were all previously 3.0 T MRI-naïve, but may not be new presentations as, because of the guidelines and increased availability of MRI, there has been added impetus to investigate the backlog of patients that have not previously been imaged.

Abnormalities related to seizure activity

In this study, the rate of abnormalities likely to be related to seizure activity was 20.2% and appears to be similar to the two largest studies published previously. Here, the positive rates were 17% in 154 patients [8] and 47% in 199 patients [12] with focal seizures, but not all patients were imaged with MRI. The detection rate is clearly higher if you only image patients with refractory epilepsy, with two studies publishing rates of 82% and 86%, respectively [14,15]. However, the NICE guidelines suggest all patients with localised epilepsy should be imaged regardless of pharmacological seizure control. From an imaging point of view, we do not make any distinction between those patients with pharmacologically controlled seizures and those with resistant symptoms. In our review, the detection rate for abnormalities was broadly similar for all age groups, though slightly higher in patients aged 30–70 years (23.4%) than in those aged <30 years (15%) or >70 years (11%). The cause for this is not known.

The detection rate of structural abnormalities potentially amenable to surgery was 15%. At our institute, patients suitable for surgery are entered into a programme of monitoring, and surgery is considered based on coexisting morbidity, pharmacological seizure control and patient wishes. During the time period studied, 49/301 (16%) patients with a structural lesion identified on MRI underwent surgical resection. An additional two patients had a normal MRI examination and surgery was based upon SPECT imaging. However, this does not reflect the surgical output from our unit as in the same period a total of 79 patients underwent epilepsy surgery. The 28 patients who underwent resection without imaging at 3.0 T were imaged elsewhere, imaged at 1.5 T (either because of protocol error or prior to 3.0 T facilities being available in our unit) or imaged outside of the study period. Many of the patients with surgically amenable pathology are in the evaluation programme and are candidates for resection in the future.

The increased use of 3.0 T MRI in this arena is likely to increase the yield of significant pathology [16-20]. However, even with the use of a high-resolution epilepsy protocol, including relaxometry of the hippocampus, almost 80% of patients in this series show no structural cause for seizure activity. It has been shown that the sensitivity can be increased slightly by using additional specialised sequences such as tractography and spectroscopy [21-23], but the small improvement in the detection rate of surgically amenable lesions may not justify the cost and extra scan time. It is more likely to be of benefit in those patients with refractory epilepsy where surgery is being considered as a treatment option.

Relaxometry and quantitative analysis can also be used to assess for bilateral MTS, although the rate detected in this series of 4.7% is lower than that reported elsewhere [24]. This suggests it is under-reported and this may have implications in the decision making process for surgical resection as there is likely to be an increased relapse rate in those with bilateral MTS [25].

Within the NICE guidelines, the role of MRI in older patients presenting with seizures is less clear. CT may be appropriate as a first-line investigation in this age group as it is sensitive to the common pathologies that cause seizures (e.g. tumours, infarcts and acute haemorrhage). However, this review shows that the prevalence of pathology amenable to surgery is similar throughout the age ranges, including those aged over 50 years (Figure 3). It is also worth noting that 25% of those patients undergoing resection for MTS in our unit were over 50 years old. While we appreciate that acute pathology may need to be ruled out in older patients, these data suggest there is value in using MRI for all patients with new onset focal seizures, especially if surgery is being considered.

Abnormalities unrelated to seizure activity

The overall prevalence of abnormal findings thought to be unrelated to seizures was 16.2% with a higher prevalence in older patients (Figure 2). Overall, this rate is much higher than our own experience of healthy volunteers sourced from a similar geographical population (8.8%) [26]. In comparison with a recent meta-analysis of 19 559 volunteers, the prevalence of aneurysms (0.25% vs 0.35%), meningiomas (0.5% vs 0.29%), pituitary tumours (0.25% vs 0.15%), Chiari malformations (0.1% vs 0.24%), epidermoids (0.05 vs 0.03%) and arachnoid cysts (0.3% vs 0.5%) did not differ dramatically [27]. However, the prevalence of cerebrovascular ischaemic disease seen in this study was high (Table 4). 178/2000 (8.9%) patients had evidence of either established infarcts or definite ischaemic change with a further 56 (2.8%) having signs that may be attributable to cerebrovascular ischaemia (33 with cerebral atrophy and 23 with non-specific white matter lesions). In comparison with our own experience of volunteers (Figure 4), ischaemic findings were unexpectedly common in younger age groups, with 53/1540 (3.7%) in those aged under 50 (p<0.0002) and 148/1946 (3.8%) in those aged under 70 years (p<0.0001). This is also higher than the prevalence reported elsewhere [27-29]. While we appreciate that by definition our study population are not comparable with healthy volunteers, these findings are currently reported as unrelated to seizure activity, and this conclusion may require revision and merit further investigation.

Figure 4.

The prevalence of changes secondary to cerebral ischaemia found in this study compared with our experience of healthy volunteers.

Conclusion

In 2005 we published the data from the first 120 patients imaged at 3.0 T for localisation-related epilepsy [29]. We discussed the potential effects of the recently produced NICE guidelines, both in terms of the demands on the neuroimaging service and the detection rate of significant pathology. Since then, the numbers of patients being referred for imaging have steadily increased from 183 in 2005 to 334 in 2007 and 410 in 2010. The rate of significant pathology has not altered and the overall detection rate of 20.2% in 2000 patients reinforces the value of prompt referral for MRI for all patients with focal seizures. This is particularly important in patients where surgery is being considered as 15% have structural lesions that may be amenable to resection. The role of neuroimaging for patients in whom surgery is not being considered is less clear. The prevalence of ischaemic disease in patients referred with localisation-related epilepsy is much higher than that seen in the general population, and these “incidental” findings may not be unrelated to seizure activity.