Pi

Ha

b

c

d

e

a

ARRAA

KLCR

1

1deedesaovgd

S

0h

Clinical Neurology and Neurosurgery 115 (2013) 591– 596

Contents lists available at SciVerse ScienceDirect

Clinical Neurology and Neurosurgery

journa l h omepage: www.elsev ier .com/ locate /c l ineuro

revalence and subtypes of radiological cerebrovascular disease in late-onsetsolated seizures and epilepsy

. Maxwell a, M. Hanbya, L.M. Parkesb, L.M. Gibsonc, C. Coutinhod, H.C.A. Emsleya,e,∗

Department of Neurology, Royal Preston Hospital, UKImaging Sciences Research Group, Biomedical Imaging Institute, University of Manchester, UKDivision of Clinical Neurosciences, Western General Hospital, Edinburgh, UKImaging Directorate, Royal Preston Hospital, UKSchool of Medicine, University of Manchester, UK

r t i c l e i n f o

rticle history:eceived 24 January 2012eceived in revised form 29 June 2012ccepted 9 July 2012vailable online 26 July 2012

eywords:ate-onset epilepsyerebrovascular diseaseadiological

a b s t r a c t

Background: Late-onset epilepsy (LOE), onset over 60, is often attributed to cerebrovascular disease (CVD),and is associated with increased stroke risk. We investigated the radiological prevalence of CVD in LOE.Methods: We undertook a retrospective case-control study of patients with LOE and age and sex-matchedcontrols, also matched for imaging modality. Radiological CVD was recorded, with radiological findingsby an experienced consultant neuroradiologist usi a structured proforma.Results: 105 cases and 105 controls were studied, comprising 61 (58.1%) males, mean (±SD) age (years)72.7 ± 7.48 (cases), 72.4 ± 7.02 (controls). 9 cases had isolated seizures rather than LOE. Imaging modality(in cases and controls) was CT in 59 and MRI in 46. Radiological CVD was more prevalent amongst cases(65.7%) than controls (33.3%) (p < 0.0001, Chi-square), odds ratio 3.83 (95% CI 2.16–6.79). Large vesseldisease (LVD) (single or multiple cortical or subcortical infarcts > 1.5 cm) was present in 23 (21.9%) casesand 2 (1.9%) (p < 0.001) controls, with small vessel disease (SVD) (periventricular or subcortical white

matter lesions (WMLs), including leukoaraiosis (LA)) present in 52 (49.5%) cases (LA in 4) and 34 (32.3%)controls (LA in 0) (p < 0.05). When WMLs were rated using a semiquantitative visual rating scale, a trendtowards greater severity was observed amongst cases compared to controls.Conclusions: Radiological CVD is significantly more prevalent in patients with LOE than controls, includingsigns of both LVD and SVD. SVD also appears to be of greater severity. Further studies are needed in this area.

. Introduction

The overall incidence of epilepsy is approximately 50 per00,000 per year in industrialised countries, although there is evi-ence of a decrease in children and a simultaneous increase in thelderly in recent decades [1]. Late-onset epilepsy (LOE, defined aspilepsy first occurring after age 60) accounts for a third of all inci-ent epilepsy requiring treatment. The traditional view was thatpilepsy in the elderly was usually due to a cerebral tumour. Foreveral decades cerebrovascular disease (CVD) has been recogniseds the commonest identified cause of LOE. The increased incidencef LOE in the elderly may relate to improved survival among indi-

iduals with CVD; but this increase seems paradoxical at first glanceiven the apparent reduction in CVD incidence over the past twoecades. However, overt clinical stroke probably constitutes the

∗ Corresponding author at: Department of Neurology, Royal Preston Hospital,haroe Green Lane, Preston PR2 9HT, UK. Tel.: +44 01772 523895.

E-mail address: hedley.emsley@manchester.ac.uk (H.C.A. Emsley).

303-8467/$ – see front matter © 2012 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.clineuro.2012.07.009

© 2012 Elsevier B.V. All rights reserved.

majority of reported CVD and under-recognition of occult CVD maywell be a contributory factor to the apparent paradox. The probabil-ity of having epileptic seizures after a first stroke has been reportedto be 5.7% within the first year and 11.5% within 5 years [2]. Corti-cal location emerges as a consistent risk factor, although subcorticallocation is also recognised to be associated with seizures [3]. It ishowever important to recognise that LOE can present without a his-tory of overt clinical CVD, yet LOE is associated with a subsequent3-fold increased stroke risk [4].

LOE poses various challenges. Firstly, the age cut off point usedin the definition of late-onset epilepsy varies considerably in thepublished literature. A threshold of 60 years is frequently used, butcut off points of 69, 50, 40 or even 20 years are used sufficientlyoften to render published studies highly heterogeneous in respectof study populations. Secondly, epilepsy is not always a straight-forward diagnosis in the elderly. Seizures in this group are most

often of focal onset, with complex partial seizures (CPS) (or focalseizures with impairment of consciousness or awareness in therevised ILAE classification [5]) being a predominant seizure type.CPS are notoriously difficult to diagnose, with significant potential

5 y and

fataattigbmrbv

tpuirrct

2

2

icairwinbawipibwwehbtmam

2C

ihuoncs

to have experienced recurrent (two or more) seizures, with 9 (9%)patients having had an isolated seizure. LOE and isolated seizure

92 H. Maxwell et al. / Clinical Neurolog

or misdiagnosis as other entities such as transient ischaemicttacks, whilst post-ictal confusion may be misconstrued as cogni-ive decline or dementia. Such issues may cause delay in diagnosisnd should prompt referral to a neurologist/epileptologist. Thirdly,nd most pertinent to this study, there is now a widely held viewhat in patients with LOE in whom no other aetiology is identified,hat occult CVD is a frequent cause, although systematic evidences, relatively lacking. The prevailing view appears to be that aood clinical outcome in terms of seizure control is often achievedecause LOE responds relatively well to anti-epileptic drug treat-ent. However, little is known about prognosis in these patients

egarding the frequently assumed underlying CVD, and there muste substantial concern that, currently, opportunities for initiatingascular prevention strategies may often be missed.

Although seizures associated with LOE are often relatively easyo treat, the potential also exists for modification of stroke risk. Thisotential will probably not be fully realised until the prevalence ofnderlying occult CVD is known, and potential mechanisms link-

ng occult CVD to LOE are better understood. We have undertaken aetrospective case control study of the prevalence and subtypes ofadiological CVD in patients with LOE or late onset isolated seizuresompared with age, gender and imaging-modality matched con-rols.

. Methods

.1. Patients and control subjects

This retrospective case-control study, based entirely on exist-ng patient records and imaging acquired during routine clinicalare, was considered to constitute audit and not to require ethicalpproval. Furthermore, the work would not influence local practice,mpose any additional burden or risk on patients, nor in its ownight would it be likely to benefit patients or society [6]. Patientsith LOE were identified from the local Epilepsy Nurse Special-

sts’ records of patients referred by local neurologists covering oureurosciences centre and local district general hospitals, by alpha-etical search wherever possible. LOE was defined as the onset afterge 60 of recurrent (two or more) epileptic seizures, in accordanceith accepted definitions [7]. Patients in whom cerebral imag-

ng had been undertaken were then identified from computerisedatient record systems. Control subjects, matched for age (accord-

ng to 5-year age bands between 60 and 84 and a 10-year bandetween 85 and 94 years), gender and modality of cranial imaging,ere identified from patients attending general neurology clinicsith a primary diagnosis of headache. Individuals with any other

xisting neurological diagnoses that could affect their likelihood ofaving CVD were excluded as control subjects. Because the num-er of suitable control subjects with available imaging exceededhe number of patients, a random selection of controls otherwise

atched for age, gender and modality of imaging was made tochieve an equal number of controls and individual patient-controlatching.

.2. Validation of brain imaging and classification of radiologicalVD

Validation of routine cerebral imaging reports was undertakenn view of variation in reporting practices between and withinospitals from which cases and controls were derived. It wasnknown whether there may be either under- or over-reporting

f radiological signs of CVD by general radiologists compared toeuroradiologists, or between differing grades of radiologists. Aonsultant neurologist (HE), reviewed all of the available patientcans and a random selection of control scans, classifying any signs

Neurosurgery 115 (2013) 591– 596

of CVD using a standardised proforma into large vessel infarcts orsmall vessel disease (SVD), haemorrhage, and other signs of CVD,and recording any other structural abnormalities including atro-phy, also with reference to the original clinical imaging report.An experienced neuroradiologist (CC), blinded to clinical diagno-sis and the original imaging report, also reviewed a randomlyselected sample of patient and control scans using the same pro-forma. White matter lesions were rated using the global score ofthe age-related white matter changes (ARWMC) semiquantitativevisual rating scale [8]. We used this scale on account of its validityfor both CT and MRI.

2.3. Statistical analysis

Statistical analyses were performed using StatsDirect version2.7.8. Comparisons employed the Chi-square test or Fisher’s exacttest. Mann–Whitney U was used for categorical non-parametricdata. Paired t-test was used for continuous parametric data. Oddsratios were calculated where appropriate. Inter-observer relia-bility was assessed using Cohen’s kappa coefficient. Any testresulting in a p-value of less than 0.05 was considered to besignificant.

3. Results

3.1. Cases and controls

Two hundred and thirteen patients with LOE or isolated seizureswere identified. Patients for whom cerebral imaging was notavailable, or where scans had been performed prior to the intro-duction of PACS (picture archiving and communication system)and were thus not readily viewable, were excluded. Patients inwhom doubt arose as to the exact age of onset of epilepsy werealso excluded. One hundred and five patients were ultimatelyincluded in the analysis. Approximately 7000 searches using thecomputerised patient record system yielded 349 potentially suit-able control subjects, 289 of whom had undergone cranial CTor MRI scans. MRI scan protocols typically included T1, T2 andfluid attenuated inversion recovery (FLAIR) sequences. 105 con-trols were selected from this group as described above. For casesand controls, imaging modality was CT in 59 (56.2%) and MRI in 46(43.8%). Brain scans included in the analysis were acquired between2003 and 2011 for cases, and between 2007 and 2011 for con-trols. CT was plain (non-contrast) in 48 (81.4%) cases and 55 (93.2%)controls.

Both cases and controls comprised 61 (58.1%) males and 44(41.9%) females. The mean (SD) age of cases and controls was 72.7(7.5) and 72.4 (7.0) years (p = 0.38). Mean (SD) age at onset ofseizures was 69.8 (12.4) years, and the mean (SD) interval betweenseizure onset and imaging was 1.5 (2.4) years. Owing to the retro-spective nature of the study and its dependence on clinical records,identification of clinical CVD may not have been complete, but a his-tory of stroke was recorded in 12 (11.4%) cases prior to the onset ofseizures.

3.2. Seizure types

Review of available clinical records revealed 96 (91%) patients

patients were therefore included in this analysis. Seizure typesrecorded in clinical records comprised generalised (presumed sec-ondary generalised) in 34.3%, complex partial seizures in 25.7%,other focal seizure types in 16.2% and unknown in 23.8%.

H. Maxwell et al. / Clinical Neurology and Neurosurgery 115 (2013) 591– 596 593

Fig. 1. Small vessel disease changes in 3 cases. (a) 76-year-old male. Axial T2 (left) and coronal T2 FLAIR (right) images show leukoaraiosis and right parietal subcorticali er lesip

3

n(S1asa(rcowsacpddtf

nfarction; (b) 71-year-old male. Axial T2 FLAIR image shows extensive white matteriventricular caps and scattered white matter lesions.

.3. Validation of radiological findings

CT or MRI scans were validated by consultant neurologist andeuroradiologist review. Overall, scans for 96 (91.4%) cases and 5451.4%) controls were reviewed by neurologist or neuroradiologist.cans were reviewed in unselected samples of 14 (13.3%) cases and0 (9.5%) controls by both a neurologist and a neuroradiologist, with

good strength of inter-rater agreement (� = 0.75). Original clinicalcan reports only were available for 9 (8.6%) cases (scans not avail-ble to view), and data were taken from scan reports only for 5148.6%) controls. Among 50 randomly selected cases and controls,adiological CVD was found in 23 (46%) where data derived fromlinical scan reports only, and 26 (52%) where data from neurologyr neuroradiology review of scans (p = 0.55). This was in keepingith a good strength of inter-rater agreement between clinical

can report and neurologist or neuroradiologist review (� = 0.72)nd suggested no significant under or overestimation of radiologi-al CVD amongst cases or controls for whom scan review was notossible. However, for those scans where neurology or neurora-

iology reviews were undertaken, analyses were based on theseata rather than clinical scan reports. With respect to identifica-ion of radiological signs of CVD, false positive and negative ratesor the neurologist (by comparison with the neuroradiologist) were

ons; (c) 74-year-old male. Axial T2 (left) and coronal T2 FLAIR (right) images show

9.1% and 15.4% respectively. False positive and negative rates forclinical scan reports (by comparison with review by neurologist orneuroradiologist) were 8.7% and 18.5% respectively.

3.4. Radiological findings other than cerebrovascular disease

Structural abnormalities other than cerebrovascular diseaselikely to have caused seizures or epilepsy were found in 19 (18.1%)cases. These comprised tumours in 10 (9.5%), sequelae of sub-arachnoid haemorrhage in 2 (1.9%), mesial temporal sclerosis in 2(1.9%), gliosis (of unknown aetiology) in 2 (1.9%), cerebral abscessin 1 (1.0%), cerebral contusions in 1 (1.0%) and chronic subduralhaematoma in 1 (1.0%). Amongst the tumours were meningiomasin 6 (5.7%), to which epileptogenesis was attributed but not neces-sarily definite in every case. Incidental findings in two cases wereconsidered unlikely to be epileptogenic (arachnoid cyst, changesconsistent with demyelination), and in two further cases wereunrelated (scalp calcifications and sinus disease). Structural abnor-malities other than CVD amongst controls comprised tumours in

7 (6.7%) (including 3 meningiomas and 1 schwannoma), gliosis(of unknown aetiology) in 1 (1.0%), evidence of previous SAH in 2(1.9%), chronic subdural haematoma in 1 (1.0%), subdural hygromain 1 (1.0%), meningocoele in 1 (1.0%), demyelination in 1 (1.0%),

594 H. Maxwell et al. / Clinical Neurology and Neurosurgery 115 (2013) 591– 596

Table 1Prevalence of radiological CVD by subtype and imaging modality in cases and controls.

Radiological CVDsubtype

Cases Controls p-Value*

All (n = 105) MRI (n = 46) CT (n = 59) All (n = 105) MRI (n = 46) CT (n = 59)

Any cerebrovascular disease 69 (65.7%) 32 (69.6%) 37 (62.7%) 35 (33.3%) 26 (56.5%) 9 (15.2%) p < 0.001

Large vessel infarcts 23 (21.9%) 7 (15.2%) 16 (27.1%) 2 (1.9%) 1 (2.2%) 1 (1.7%) p < 0.001Cortical, single 14 (13.3%) 4 (8.7%) 10 (16.9%) 0 (0%) 0 (0%) 0 (0%)Cortical, multiple 5 (4.8%) 0 (0%) 5 (8.5%) 1 (1.0%) 1 (2.2%) 0 (0%)Subcortical, single 3 (2.9%) 2 (4.4%) 1 (1.7%) 1 (1.0%) 0 (0%) 1 (1.7%)Subcortical, multiple 1 (1.0%) 1 (2.2%) 0 (0%) 0 (0%) 0 (0%) 0 (0%)

Small vessel disease (SVD) 52 (49.5%) 30 (65.2%) 22 (37.3%) 34 (32.4%) 27 (58.7%) 7 (11.9%) p = 0.012Periventricular white matter lesions (WML) 27 (25.7%) 8 (17.4%) 19 (32.3%) 5 (4.8%) 2 (4.4%) 3 (5.1%)Subcortical WML 30 (28.6%) 28 (60.9%) 2 (3.4%) 26 (24.8%) 23 (50%) 3 (5.1%)Leukoaraiosis 4 (3.8%) 3 (6.5%) 1 (1.7%) 0 (0%) 0 (0%) 0 (0%)Lacunar infarcts ≤1.5 cm 6 (5.7%) 1 (2.2%) 5 (8.5%) 2 (1.9%) 1 (2.2%) 1 (1.7%)Enlarged perivascular spacesa 17 (16.1%) 16 (34.8%) 1 (1.7%) 9 (8.6%) 8 (17.4%) 1 (1.7%)

Haemorrhage 5 (4.8%) 1 (2.2%) 4 (6.8%) 1 (1.0%) 0 (0%) 1 (1.7%) p = 0.030Intracerebral haemorrhage 2 (1.9%) 0 (0%) 2 (3.4%) 0 (0%) 0 (0%) 0 (0%)Haemorrhagic transformation of infarct 1 (1.0%) 0 (0%) 1 (1.7%) 1 (1.0%) 0 (0%) 1 (1.7%)Microhaemorrhage 1 (1.0%) 1 (2.2%) – 0 (0%) 0 (0%) –

ss

cwGp

3

prpvp23afF2

oo(vta

CSs3ow(

lcwt

to age, gender, and imaging-modality matched controls. Large ves-sel infarcts, SVD and haemorrhage were all identified significantlymore frequently in cases than in controls. SVD was also moresevere amongst cases than controls. This was a retrospective study

Table 2Distribution of scores on ARWMC scale in cases and controls.

Score Cases (n = 105) Controls (n = 105) p-Value

Subarachnoid haemorrhage 1 (1.0%) 0 (0%)

* Comparisons between all cases and controls.a Excluded from SVD totals.

uspected skull lesion in 1 (1.0%), calcified nodules in 3 (2.9%) andinus disease in 7 (6.7%).

No significant differences were observed between cases andontrols in respect of cerebral atrophy (focal or generalised) whichas identified in 47 (44.8%) cases and 39 (37.1%) controls (p = 0.26).eneralised atrophy considered to be proportionate for age wasresent in 37 (35.2%) cases and 33 (31.4%) controls (p = 0.6).

.5. Radiological cerebrovascular disease, subtypes and severity

Radiological evidence of any cerebrovascular disease wasresent in 69 (65.7%) cases and 35 (33.3%) controls (p < 0.001; oddsatio [OR] 3.83 [95% CI 2.16–6.79]). A statistically significant higherrevalence of radiological evidence of large vessel infarcts, smallessel disease and haemorrhage was observed among cases com-ared to controls (Table 1). Large vessel infarcts were identified in3 (21.9%) cases and 2 (1.9%) controls (p < 0.001; OR 14.45 [95% CI.31–63.06]). Small vessel disease was identified in 52 (49.5%) casesnd 34 (32.4%) controls (p < 0.05; OR 2.05 [95% CI 1.17–3.59]). ORsollowing the exclusion of cases with history of clinical stroke (seeig. 1) were 7.63 (95% CI 1.66–35.06) for large vessel infarcts and.23 (95% CI 1.25–3.97) for small vessel disease.

After exclusion of the 19 cases with structural abnormalitiesther than cerebrovascular disease likely to have caused seizuresr epilepsy, evidence of small vessel disease was present in 4653.5%) cases (compared to 32.4% controls, p = 0.005), with smallessel disease alone seen in 36 (41.9%) cases and 34 (32.4%) con-rols (p = 0.17) and large vessel infarcts alone in 12 (14.0%) casesnd 1 (1.0%) control (p = 0.0006).

As would be expected, MRI exhibited greater sensitivity thanT for identification of radiological CVD, particularly in respect ofVD. Considering all 210 subjects (cases and controls), radiologicaligns of CVD were identified on 63.0% of 92 MRI scans compared to9% of 118 CT scans. Large vessel infarcts were identified on 8.7%f MRI scans and 14.4% of CT scans (p = 0.20) while evidence of SVDas identified on 62.0% of MRI scans compared to 24.6% of CT scans

p < 0.0001).WMLs in periventricular or subcortical distribution, or

eukoaraiosis (diffuse, confluent WMLs), were present in 61 (58.1%)ases and 31 (29.5%) controls overall. Scans from these subjectsere graded using the global score of the age-related white mat-

er changes (ARWMC) semiquantitative visual rating scale [8], on

1 (1.7%) 0 (0%) 0 (0%) 0 (0%)

which symmetrical, well-defined periventricular caps or bands arenot considered, with scores of between 0 and 3 assigned based onincreasing WML burden. The distribution of scores amongst casesand controls according to the ARWMC scale is shown in Table 2.A trend towards greater WML severity amongst cases compared tocontrols was observed, with a greater proportion of controls scoring0 and a greater proportion of cases scoring 2 or 3.

3.6. Radiological findings amongst cases with history of clinicalstroke

Among the 12 cases in whom review of clinical records revealeda clinical history of stroke, evidence of large vessel infarcts waspresent in 11 (91.7%), with a single cortical infarct in 8 (66.7%) andmultiple cortical infarcts in 3 (25.0%). Radiological SVD was iden-tified in 4 (33.3%), comprising periventricular WMLs in 3 (25.0%),lacunar infarct in 1 (8.3%) and enlarged perivascular spaces in 1(8.3%).

3.7. Radiological findings amongst cases with isolated seizures

Among the 9 cases with isolated seizures rather than LOE, 8(88.9%) had evidence of radiological CVD, comprising a single cor-tical infarct and WMLs in 1 (11.1%), SAH in 1 (11.1%), and SVD in 6(66.7%) (WMLs or LA in all 6).

4. Discussion

We have found a significantly greater prevalence of radiolog-ical CVD amongst cases with LOE or isolated seizures, compared

0 61 (58.1%) 75 (71.4%) 0.061 28 (26.7%) 27 (25.7%) 0.882 11 (10.5%) 3 (2.9%) 0.053 5 (4.8%) 0 (0%) 0.06

y and

uictTbtipwlwlitcpLaei

Locetedwwtasw

scpcwrscp4taoo

tSttcistafcdlcd

H. Maxwell et al. / Clinical Neurolog

sing routine clinical records and imaging, an approach with somenherent limitations. Reliable identification of clinical cerebrovas-ular disease, for example, is difficult on a retrospective basis inhe absence of disabling stroke. It is possible that minor stroke orIA may have been underestimated. Furthermore the long intervaletween seizure onset and imaging likely reflects a variety of fac-ors including delays in referral or diagnosis. The majority of casesn this series did have epilepsy, although it is interesting that therevalence of CVD, and in particular SVD, was in fact greater in casesith isolated seizures than in those with LOE. In the absence of

onger term follow up we cannot exclude the possibility that casesith isolated seizures did not go on to develop LOE. The preva-

ence of large vessel infarcts was expectedly high amongst thosendividuals with a history of clinical stroke. Structural abnormali-ies other than cerebrovascular disease, particularly tumours, had aomparable prevalence in this study to earlier work. Overall, in theresent study such abnormalities likely to have caused seizures orOE were found in 18% cases, with tumours comprising 9.5% over-ll. By comparison, in an earlier study of 81 elderly patients withpileptic seizures, tumours were identified in 12% of 60 patientsnvestigated by cerebral CT [9].

Existing evidence of the association between occult CVD andOE is relatively limited, despite the widely held view that CVDften underlies LOE. In one cross-sectional, community-based,ase-control study comprising 4944 subjects, 65 of whom hadpilepsy, a history of stroke was strongly associated with life-ime epilepsy (OR 3.3; 95% CI 1.3–8.5), as well as with late-onsetpilepsy (OR 3.1; 95% CI 0.9–10.6) [10]. History of stroke, myocar-ial infarction, peripheral vascular disease, and vascular risk factorsere significantly associated with lifetime epilepsy and LOE. Evenhen stroke patients were excluded, odds ratios remained statis-

ically significant for the relationships between total cholesterolnd left ventricular hypertrophy and LOE, suggesting a relation-hip between vascular factors and the risk of LOE, apart from thathich exists through clinically overt stroke.

The prevalence of occult CVD in patients with LOE has been theubject of very few studies. In one study, significantly more occulterebral infarcts were found amongst patients with single or multi-le seizures starting after the age of 40 compared with age matchedontrols, with the prevalence of occult CVD on CT amongst patientsith onset over 60 with LOE being 21.4% [11]. By contrast, a ret-

ospective analysis of medical notes from patients aged over 50 ateizure onset identified only 1 out of 130 patients had a ‘cerebrovas-ular focus’ without clinical stroke [12]. However, in a study of 4709atients without history of CVD developing seizures after 60 and709 matched controls, the risk of subsequent clinical stroke afterhe development of late onset seizures was found to be increasedlmost 3-fold (OR 2.89, 95% CI 2.45–3.41) [4]. Heralding seizuresr history of LOE prior to stroke has been described in at least twother studies [13,14].

Epilepsy is usually regarded as deriving from the cerebral cor-ex. The increased prevalence of largely subcortical radiologicalVD changes in cases compared to controls – and indeed the trendowards greater WML severity – is particularly interesting. Afterhe exclusion of cases ultimately found to have evidence of clini-al stroke, the odds ratio for (predominantly cortical) large vesselnfarcts reduced whilst that for small vessel disease increasedlightly. The prevailing view is that CVD often underlies LOE. Forhere to be a causal relationship between CVD and LOE in thebsence of apparent cortical lesions then we might consider theollowing possibilities. Firstly, an increased burden of subcorti-al SVD visible on conventional imaging might be associated with

isruption of corticosubcortical circuits and consequently under-

ie epileptogenicity. Secondly, the herein reported radiologicalhanges of SVD may merely indicate the presence of other changesue to cerebrovascular disease not identified using conventional

Neurosurgery 115 (2013) 591– 596 595

imaging, for example, cortical microinfarcts, microhaemorrhages,disrupted white matter integrity, increased blood–brain barrierpermeability, or other changes associated with disruption of neu-rovascular unit integrity such as disruption of cerebral metabolismand/or perfusion, many of which might be considered to promoteepileptogenicity [15]. Increased blood–brain barrier permeabilityin particular is increasingly recognised in the context of SVD [16].

Some of these abnormalities associated with SVD are cur-rently only identifiable histopathologically [17]. Furthermore, evenwidely accepted MRI ‘visible’ expressions of SVD such as WMLs canshow relatively weak clinicoradiological association, likely partlydue to the heterogeneous pathological substrates of WMLs. Morerecently, the spectrum of ‘visible’ SVD markers has been expanding,and is considered to include enlarged perivascular spaces (EPVS)[18] and brain volume [19], at least as a surrogate marker. Previousstudies reporting radiological CVD in LOE have largely not includedmore recently recognised markers of CVD. In the present, we didobserve an increased number of EPVS amongst cases comparedto controls, although atrophy assessed non-quantitatively did notdiffer significantly between cases and controls.

5. Conclusion

We have demonstrated an increased prevalence of radiolog-ical CVD in patients with LOE and isolated late onset seizures.We observed large vessel infarcts, SVD and haemorrhage signifi-cantly more frequently in cases than in controls. WMLs showedgreater severity in cases than in controls. Any causal relationshipbetween SVD and epileptogenicity remains, at this stage, specula-tive. Mechanistic links between occult CVD, and in particular SVD,and epileptogenicity are currently unclear and there is no doubtthat further studies are required in this area.

Funding

MH, LMG, CC and HCAE are funded by the NHS; LMP is fundedby the University of Manchester.

Conflict of interest

The authors have no competing interests to declare.

Acknowledgements

We are grateful for the help of Lesley North and StephanieDixon, Epilepsy Specialist Nurses, Royal Preston Hospital, for theirassistance; and for statistical advice from the Department of Math-ematics and Statistics, University of Lancaster.

References

[1] Hauser WA, Annegers JF, Kurland LT. Incidence of epilepsy and unprovokedseizures in Rochester, Minnesota: 1935–1984. Epilepsia 1993;34(3):453–68.

[2] Burn J, Dennis M, Bamford J, Sandercock P, Wade D, Warlow C. Epilepticseizures after a first stroke: the Oxfordshire Community Stroke Project. BMJ1997;315(7122):1582–7.

[3] Camilo O, Goldstein LB. Seizures and epilepsy after ischemic stroke. Stroke2004;35:1769–75.

[4] Cleary P, Shorvon S, Tallis R. Late-onset seizures as a predictor of subsequentstroke. Lancet 2004;363(9416):1184–6.

[5] Berg AT, Berkovic SF, Brodie MJ, Buchhalter J, Cross JH, van Emde Boas W, et al.Revised terminology and concepts for organization of seizures and epilepsies:report of the ILAE Commission on Classification and Terminology, 2005–2009.Epilepsia 2010;51:676–85.

[6] Wade D. Ethics audit and research: all shades of grey. BMJ 2005;330:468–71.

[7] ILAE. Proposal for revised classification of epilepsies and epileptic syndromes.

Commission on classification and terminology of the international leagueagainst epilepsy. Epilepsia 1989;30:389–99.

[8] Wahlund LO, Barkhof F, Fazekas F, Bronge L, Augustin M, Sjögren M, et al.European task force on age-related white matter changes. A new rating

5 y and

[

[

[

[

[

[

[

[

[

96 H. Maxwell et al. / Clinical Neurolog

scale for age-related white matter changes applicable to MRI and CT. Stroke2001;32(6):1318–22.

[9] Roberts MA, Godfrey JW, Caird FI. Epileptic seizures in the elderly. I. Aetiologyand type of seizure. Age and Ageing 1982;11(1):24–8.

10] Li X, Breteler MM, de Bruyne MC, Meinardi H, Hauser WA, HofmanA. Vascular determinants of epilepsy: the Rotterdam Study. Epilepsia1997;38(11):1216–20.

11] Roberts RC, Shorvon SD, Cox TC, Gilliatt RW. Clinically unsuspected cerebralinfarction revealed by computed tomography scanning in late onset epilepsy.Epilepsia 1988;29(2):190–4.

12] Paradowski B, Zagrajek MM. Epilepsy in middle-aged and elderly people: a

three-year observation. Epileptic Disorders 2005;7(2):91–5.

13] Daniele O, Mattaliano A, Tassinari CA, Natalè E. Epileptic seizures and cere-brovascular disease. Acta Neurologica Scandinavica 1989;80(1):17–22.

14] Shinton RA, Gill JS, Zezulka AV, Beevers DG. The frequency of epilepsy precedingstroke. Case-control study in 230 patients. Lancet 1987;1(8523):11–3.

[

Neurosurgery 115 (2013) 591– 596

15] Gibson LM, Allan SM, Parkes LM, Emsley HC. Occult cerebrovascular diseaseand late-onset epilepsy: could loss of neurovascular unit integrity be a usefulmodel? Cardiovascular Psychiatry and Neurology 2011;2011:130406.

16] Wardlaw JM. Blood–brain barrier cerebral small vessel disease. Journal of theNeurological Sciences 2010;299(1–2):66–71.

17] Gouw AA, Seewann A, van der Flier WM, Barkhof F, Rozemuller AM, Schel-tens P, et al. Heterogeneity of small vessel disease: a systematic review ofMRI and histopathology correlations. Journal of Neurology, Neurosurgery andPsychiatry 2011;82(2):126–35.

18] Zhu YC, Tzourio C, Soumaré A, Mazoyer B, Dufouil C, Chabriat H. Sever-ity of dilated Virchow–Robin spaces is associated with age, blood pressure,

and MRI markers of small vessel disease: a population-based study. Stroke2010;41(11):2483–90.

19] Nitkunan A, Lanfranconi S, Charlton RA, Barrick TR, Markus HS. Brain atro-phy and cerebral small vessel disease: a prospective follow-up study. Stroke2011;42(1):133–8.