doi:10.1053/seiz.2001.0613, available online at http://www.idealibrary.com onSeizure 2002; 11: 340–343

MRI evidence of mesial temporal sclerosis in subjectswithout seizures

SELIM R. BENBADIS†‡¶‖

, JENNIFER WALLACE§, F. REED MURTAGH§, CARLOS MARTINEZ§,WILLIAM O. TATUM, IV† & FERNANDO L. VALE‡

Departments of †Neurology, ‡Neurological Surgery, and §Radiology, University of South Florida andTampa General Hospital, Comprehensive Epilepsy Program; ¶The James A. Haley VeteransAdministration Hospital, Tampa, FL, USA

Correspondence to: Selim R. Benbadis, MD, Associate Professor, Departments of Neurology & Neurosurgery,University of South Florida College of Medicine, 4 Columbia Drive, Suite 730, Tampa, FL 33606, USA.E-mail: sbenbadi@hsc.usf.edu

The identification of mesial temporal sclerosis (MTS) on magnetic resonance imaging (MRI) is critical in the evaluation fortemporal lobectomy. The purpose of this study was to estimate the prevalence of MTS by MRI in the general population.We reviewed the charts of patients undergoing MRI scans over an 18-month period. In order to be included, patients had tohave no history of seizures, and the MRIs had to read as ‘normal’. Patients were excluded if there was a history of seizures,mental illness, or alcohol abuse, and if they had a family history of seizures. Ages had to be from 5 to 50 years. All patientshad MRI scans performed in three planes on 1.5 T systems. MRI diagnosis of MTS was based on standard visual diagnosticcriteria analyzing: (1) hippocampal atrophy and (2) increased hippocampal T2 signal. Out of 375 charts reviewed, 204 met theinclusion criteria. Of the 204, 29 (14%) had abnormal and asymmetric mesial temporal structures consistent with a diagnosisof unilateral MTS. Nineteen (9.3%) had hippocampal atrophy only, two (1%) had T2 signal abnormalities, and eight (4%) metboth criteria. There are some ‘false positives’ and MRI evidence for MTS is only meaningful in the proper clinical context.

c© 2002 Published by Elsevier Science Ltd on behalf of BEA Trading Ltd.

Key words:seizures; MRI; mesial temporal sclerosis; neuroimaging.

INTRODUCTION

Mesial temporal sclerosis (MTS) is the most commoncause of temporal lobe epilepsy, and magnetic reso-nance imaging (MRI) evidence for MTS is critical inthe evaluation for temporal lobectomy. First, unilateralhippocampal atrophy predicts a favorable surgicaloutcome1–3. For example, when there is concordancebetween ictalelectroencephalogram (EEG) and hip-pocampal atrophy, surgical outcome is excellent ingreater than 95% of cases4, 5. In fact, the combinedMRI andEEG findings have become so important that,in some cases, temporal lobectomy may be consideredwithout ictal recordings6–9.

We10 recentlydescribed a small series of patientswith psychogenicnon-epileptic seizures whose MRIshowed evidence for MTS. Thus, while the sensitivity

and usefulness of MRI for MTS is unquestionablyhigh, its specificity (i.e. rate of ‘false positives’) isunknown. The purpose of this study was to estimatethe frequency of such abnormalities in subjectswithout seizures.

METHODS

Study population

We reviewed the charts of patients undergoing MRIscans over an 18-month period. In order to beincluded, patients had to meet the following criteria:

• Ages from 5 to 50 years.

• No history of seizures, psychiatric disease, andalcohol abuse.

‖

http://hsc.usf.edu/~sbenbadi

1059–1311/02/$22.00/0 c© 2002 Published by Elsevier Science Ltd on behalf of BEA Trading Ltd.

MRI evidence of mesial temporal sclerosis 341

• No family history of seizures.

MRI had to meet the following criteria:

• Performed for vague non-specific symptoms(headaches, dizziness) or screening for metastasesor cerebral aneurysms.

• Read as ‘normal’ during the routine (and official)interpretation, not specifically examining mesialtemporal structures.

MRI technique

All patients had MRI scans performed in three planeson 1.5 T systems, with slice thickness of 4 mm, andall scans afforded adequate visualization of the mesialtemporal lobe structures.

MRI review

These MRI scans were then examined by a neu-roradiologist (FRM, senior member of the ASNRand Certificate of added qualification). Estimates ofhippocampal volume were made on both T1 andT2-weighted images in all patients. MRI diagnosisof MTS was based on standard visual diagnosticcriteria1–5 analyzing: (1) hippocampal atrophy and(2) increased hippocampal T2 signal.

RESULTS

Results are summarized in Fig. 1. We reviewed 375charts, and 204 met the inclusion criteria. Of the204, 29 (14%) had abnormal and asymmetric mesialtemporal structures consistent with a diagnosis ofunilateral MTS. Nineteen (9.3%) had hippocampalatrophy only, two (1%) had T2 signal abnormalities,and eight (4%) met both criteria. In addition, another32 (15.7%) patients were thought to have symmetricmesial structures abnormalities. An example of normaland hippocampal atrophy (right side) are shown (seeFigs. 2 and 3).

DISCUSSION

In this series, we found that 14% of patients under-going MRI for ‘non-seizure reasons’, and whose MRIfailed to disclose gross abnormalities, had asymmetricabnormalities of the mesial temporal structures thatcould suggest a diagnosis of MTS. In fact, 4% of thepatients had reasonably definite evidence for MTS,meeting both criteria.

Charts reviewed(n = 375)

MRIs reviewed MRIs eliminatedn = 204 171

Normal Abnormal143 (70%) 61 (30%)

Asymmetric Symmetric29 (14.2%) 32 (15.7%)

Hippocampal atrophy T2 signal change Both criteria19 (9.3%) 2 (1%) 8 (4%)

Fig. 1: Distribution of the patients.

Fig. 2: Normal T2 weighted coronal sections.

Fig. 3: T2 weighted coronal sections—example ofhippocampal atrophy.

That MRI evidence that MTS can be present inpatients without seizures is not all that surprising,

342 S. R. Benbadis et al.

since even in the best series 10–20% of patientswith clear hippocampal atrophy can fail a temporallobectomy, possibly indicating that the MRI abnor-mality was not the source of the seizures. Our rateof ‘false positive’ is consistent with the specificity of86% reported by others11. Hippocampal atrophy hasalso beenreported in children with benign childhoodepilepsy with centrotemporal spikes12, where it islikely to represent an incidental finding. Thus, whileMRI evidence of MTS is critical in thepresurgicalevaluation, it should not be used to make a diagnosisof epilepsy. Indeed, MRI evidence of MTS only hasmeaning in the proper (electro-clinical) context10, 13.This pitfall is in many ways comparable to whathappened with the diagnosis of multiple sclerosis(MS) in the early days of MRI. Because of theextreme sensitivity of MRI, many patients were givenpresumptive diagnoses of MS based on MRI findings,until it became apparent that these findings were non-specific and that many subjects without symptomshad MRI ‘consistent with MS’. Another analogousexample is the specificity of MRI for lumbosacralspondylosis14, 15. To some degree a comparablesituation exists for most imaging or laboratory tests.As with the MRI of MS, the MRI of MTS is only ofvalue when interpreted in the proper clinical context.

The prevalence of such mesial temporal abnor-malities in subjects without seizures has not beenextensively studied. One series16 found such ‘falsepositives’ in only two (1%) of 207 patients who had anMRI for hearing loss, but these were performed usingonly T2-weighted fast spin-echo imaging. Because ofits high sensitivity, MRI often reveals other incidentalabnormalities even in healthy volunteers, includingwhite matter lesions, old lacunar infarcts, arachnoidcysts, and cavernous angiomas17. Compared to otherabnormalities, MTSis an important one becauseseizures are frequent (affecting at least 1% of thepopulation), and MTS in the presence of seizures mayindicate that seizures are likely to be intractable18 andcould betreated with surgery.

Several potential shortcomings of our study warrantdiscussion. First, hippocampal atrophy has beenreported in dementia with Lewy body19, alcoholism20

and schizophrenia21. For that reason, we were carefulto exclude older patients, and patients with a historyof dementia, psychiatric disease, or alcohol abuse.However, we cannot absolutely exclude that ourpatients with MRI evidence of MTS later developedany of these conditions, or even temporal lobeseizures. Second, our sample did not truly representa ‘normal’ population. Nevertheless, we feel that itwas an adequate approximation, since the symptomsor reasons for the MRIs were clearly irrelevant tohippocampal atrophy. Third, using different imagingprotocols may produce different results than the

routine technique used here. For example, quantitativeanalysis or ‘volumetry’4 and FLAIR images22 mayimprove sensitivity of MRI for the detection ofMTS. We also recognize that there is an elementof subjectivity when visually analyzing MRI forthe possibility of MTS. For that reason, we usedvarious criteria with two degrees of ‘evidence’ (onecriterion or two criteria). Thus, our findings are stilluseful since we are at least getting a minimalistic(conservative) estimate (underestimate) of the ‘falsepositive’ MRI. Our findings would indicate thatatleast 4% of the population may have MRI evidencefor unilateral hippocampal atrophy, documenting thepossibility of ‘false positives’. Ideally, our findingshould be confirmed by a prospective study inhealthy individuals.

REFERENCES

1. Kuzniecky, R., Burgard, S., Faught, E., Morawetz, R. andBartolucci, A. Predictive value of MRI in temporal lobeepilepsy surgery.Archives of Neurology1993;50: 65–69.

2. Berkovic, S. F., McIntosh, A. M., Kalnins, R. M., Jack-son, G.D., Fabinyi, G. C., Brazenor, G. A., Bladin, P. F. andHopper, J. L. Preoperative MRI predicts outcome of temporallobectomy: an actuarial analysis.Neurology 1995; 45:1358–1363.

3. Kuzniecky, R., Cascino, G. D., Palmini, A.et al. Structuralimaging.In: Surgical Treatment of the Epilepsies. 2nd Edition(Ed. J. Engel Jr). New York, Raven Press, 1993: pp. 197–200.

4. Jack, C. R., Sharborough, F. W., Cascino, G. D., Hirschorn,K. A., O’Brien, P. C. and Marsh, W. R. MRI-based hip-pocampal volumetry: correlation with outcome after temporallobectomy.Annals of Neurology1992;31: 138–146.

5. Garcia, P. A., Laxer, K. D., Barbaro, N. M. and Dil-lon, W. P. Prognostic value of qualitative magnetic reso-nance imaging hippocampal abnormalities in patients under-going temporal lobectomy for medically refractory seizures.Epilepsia1994;35: 520–524.

6. Cascino, G. D., Trenerry, M. R., So, E. L.et al. RoutineEEG and temporal lobe epilepsy: relation to long-termEEG monitoring, quantitative MRI, and operative outcome.Epilepsia1996;377: 651–656.

7. Pataraia, E., Lurger, S., Serles, W.et al. Ictal scalp EEG inunilateral mesial temporal lobe epilepsy.Epilepsia1998;39:608–614.

8. Holmes, M. D., Dodrill, C. B., Ojemann, L. M. andOjemann, G.A. Five-year outcome after epilepsy surgery innonmonitored and monitored surgical candidates.Epilepsia1996;37: 748–752.

9. Cendes, F., Li, L. M., Watson, C., Andermann, F.,Dubeau, F. and Arnold, D. L. Is ictal recording mandatoryin temporal lobe epilepsy? Not when the interictal electroen-cephalogram and hippocampal atrophy coincide.Archives ofNeurology 2000;57: 497–500.

10. Benbadis, S. R., Tatum, W. O IV., Vale, F. andMurtagh, F. R. MRI-evidence of mesial temporal sclerosis inpatients with psychogenic nonepileptic seizures.Neurology2000;55: 1061–1062.

11. Jackson, G. D., Berkovic, S. F., Tress, B. M., Kalnins, R. M.,Fabinyi, G. C. and Bladin, P. F. Hippocampal sclerosis can bereliably detected by magnetic resonance imaging.Neurology1990;40: 1869–1875.

12. Lundberg, S., Eeg-Olofsson, O., Raininko, R. and Eeg-

MRI evidence of mesial temporal sclerosis 343

Olofsson, K. E. Hippocampal asymmetries and white matterabnormalities onMRI in benign childhood epilepsy withcentrotemporal spikes.Epilepsia1999;40: 1808–1815.

13. Spencer, S. S. MRI and epilepsy surgery.Neurology1995;45:1248–1250.

14. Jensen, M. C., Brant-Zawadski, M. N., Obuchowski, N.et al. MRI of the lumbar spine in people without back pain.New England Journal of Medicine1994;331: 69–73.

15. Boden, S. D., Davis, D. O., Dina, T. S., Patronas, N. J. andWiesel, S. W. Abnormal magnetic-resonance scans of thelumbar spine in asymptomatic subjects. A prospectiveinvestigation.Journal of Bone and Joint Surgery. AmericanVolume1990;72: 403–408.

16. Moore, K. R., Swallow, C. E. and Tsuruda, J. S. Incidentaldetection of hippocampal sclerosis on MR images: is itsignificant?American Journal of Neuroradiology1999; 20:1609–1612.

17. Katzman, G. L., Dagher, A. P. and Patronas, N. J. Incidentalfindings on brain magnetic resonance imaging from 1000asymptomatic volunteers.JAMA1999;282: 36–39.

18. Semah, F., Picot, M. C., Adam, C.et al. Is the underlyingcause of epilepsy a major prognostic factor for recurrence?Neurology1998;51: 1256–1262.

19. Barber, R., Ballard, C., McKeith, I. G., Gholkar, A. andO’Brien, J. T. MRI volumetric study of dementia withLewy bodies: a comparison with AD and vascular dementia.Neurology2000;54: 1304–1309.

20. Agartz, I., Momenan, R., Rawlings, R. R., Kerich, M. J. andHommer, D. W. Hippocampal volume in patients withalcohol dependence.Archives of Genetic Psychiatry1999;56:356–363.

21. Shenton, M. E., Kikinis, R., Jollesz, F. A.et al.Abnormalitiesof the left temporal lobe and thought disorder in schizophrenia.A quantitative magnetic resonance imaging study.NewEngland Journal of Medicine1992;327: 604–612.

22. Bergin, P. S., Fish, D. R., Shorvon, S. D.et al. Magneticresonanceimaging in partial epilepsy: additional abnormalitiesshown with the fluid attenuated inversion recovery (FLAIR)sequence.Journal of Neurology, Neurosurgery and Psychiatry1995;58: 439–443.