Epilepsia, 36(9):905-910, 1995 Lippincott-Raven Publishers, Philadelphia 0 International League Against Epilepsy

Bilateral Hippocampal Atrophy in Medial Temporal Lobe Epilepsy

David King, Susan S. Spencer, *Gregory McCarthy, "Marie Luby, and "Dennis D. Spencer

Departments of Neurology and *Surgery, Yale University School of Medicine, New Haven, and VA Medical Center, West Haven, Connecticut, U.S.A.

Summary: Quantitative evidence of hippocampal atrophy has been correlated with site of seizure onset, hippocam- pal neuronal loss, and seizure relief after resection. Most studies have quantified hippocampal atrophy using ratios or differences between right and left hippocampal values. However, bilateral hippocampal atrophy may remain un- detected by these techniques. To assess the frequency and implications of bilateral hippocampal atrophy, we studied absolute hippocampal volumes in 53 temporal lobectomy patients who had undergone intracranial elec- troencephalogram recordings preoperatively . Coronal im- ages were constructed perpendicular to the longitudinal axis of the hippocampus. Atrophy was defined as >2 SD below control values in the volume of the posterior 1.5 cm of the hippocampus. Five of 53 patients (9%) had bilateral

Hippocampal atrophy detected by quantitative magnetic resonance imaging (MRI) correlates with the presence of mesial temporal sclerosis (MTS), the most common pathological lesion in patients with medial temporal lobe epilepsy (MTLE) (1). As a result of this fact, quantitative MRI is increasingly performed in the presurgical evaluation of patients with uncontrolled partial epilepsy. Documentation of unilateral hippocampal atrophy with concordant scalp electroencephalogram (EEG) localization of- ten allows surgical treatment without intracranial EEG (2). MRI volumetry using ratios or volume differences between left and right hippocampus is highly sensitive to unilateral atrophy, but it may fail to detect patients with bilateral disease (3-5).

The subgroup of patients with MTLE and bilat- eral, independent, medial temporal lobe seizure on- sets represents a diagnostic and therapeutic chal-

Received November 29, 1994; revision accepted March 8, 1995.

Address correspondence and reprint requests to Dr. S. S. Spencer at Department of Neurology, Yale University School of Medicine, PO Box 208018, New Haven, CT 06520-8018, U.S.A.

hippocampal atrophy; four of these cases were undetec- ted by ratios. Surgery was performed on the side of ictal onset in all five patients; four have been seizure-free for >2 years. These results suggest that (a) mesial temporal sclerosis can be present bilaterally and may go undetec- ted by hippocampal ratio or difference measures; (b) ab- solute hippocampal volume values as well as ratios are needed to detect all patients with bilateral hippocampal atrophy; and (c) temporal lobectomy is not contraindi- cated in patients with bilateral hippocampal atrophy, but success depends on electroencephalographic documenta- tion of the side of predominant ictal onset. Key Words: Hippocampal atrophy-Quantitative MRI-Temporal lobe epilepsy-Mesial temporal sclerosis-Depth elec- trodes.

lenge. Hirsch et al. (6) demonstrated that it is pos- sible to obtain excellent outcomes from temporal lobectomy in this group of patients. Spencer et al. (3) showed found only one third of these patients demonstrated unilateral hippocampal atrophy on MRI when ratios were used. This low incidence of hippocampal atrophy could be explained by the fail- ure of ratios and difference measures to detect bi- lateral atrophy of varying degrees.

We hypothesized that the use of absolute hippo- campal volumes would improve the sensitivity of quantitative MRI to all degrees of unilateral and bilateral hippocampal atrophy. This improved sen- sitivity could enable identification of additional pa- tients with hippocampal atrophy and also provide a measure of the prevalence of bilateral atrophy, a presumptive marker for bilateral MTS.

METHODS

Patients We studied all medically refractory MTLE pa-

tients in the Yale epilepsy program database who

905

906 D . KING ET AL.

had absolute volumetric MRI of hippocampal size and intracranial EEG studies for localization of sei- zure onset before surgical resection. Patients were studied and treated between 1988 and 1993. Details of the initial evaluation of patients considered for surgical treatment have been published elsewhere (7). Patients had various combinations of subdural strips and depth electrodes inserted to sample the medial temporal structures bilaterally. Placement was confirmed with postimplantation MRI (8). Recording

Patients were monitored with continuous audio- visual-EEG recordings obtained using a 64-channel Telefactor Beehive System. Anticonvulsant drugs were tapered and withdrawn until at least three typ- ical, spontaneous seizures were recorded. EEG re- cordings were reviewed on reformatted hard copy by three board-certified electroencephalographers with experience in intracranial EEG monitoring.

Patients were classified as having unilateral onset when all of the recorded spontaneous seizures be- gan in the same medial temporal lobe structures. Patients were considered to have bilateral onset when seizures began independently in both left and right medial temporal lobes (in any proportion). In- terictal abnormalities were not used for this classi- fication. MRI

MRI volumetric measurements were obtained from coronal images using an SPGR (spoiled gradi- ent-echo) imaging sequence [relaxation time (TRY excitation time (TE), 24/5 ms) with a 16-cm field of view and 3-mm-thick slices. The images were highly T1-weighted and showed excellent gray-white mat- ter contrast. Before measurement, the images were assessed for rotation around the vertical axis by comparing left-right (L-R) symmetry of the internal auditory canals (IACs) and the anterior commis- sure. Rotation was corrected by interpolating the entire image series onto a new L-R axis defined by two structurally symmetric points, such as the IACs. The plane of the hippocampi was then mea- sured by defining three points: the centers of the left and right hippocampi in the image 3 mm anterior to the colliculi and the center of the right or left hip- pocampus in the image 15 mm anterior to the colli- culi. An oblique “coronal” series perpendicular to this plane was then interpolated. Measurements on this new series were thus uniformly made orthogo- nal to the long axis of the hippocampi. The hippo- campi were outlined by cursor in the five images 15 mm anterior to the colliculi, which included the mid-body but excluded the pes. This approach was used to avoid difficulty in discriminating the amygdala from hippocampus anteriorly. The hippo-

campal volume was calculated by adding all the pix- els and multiplying the total by the pixel size and slice thickness.

Control values were obtained from 17 normal men who had a mean age of 39 years. The mean left hippocampal volume was 1.28 cc (SD, 0.11 cc); the mean right hippocampal volume was 1.28 cc (SD, 0.12 cc). Hippocampal atrophy was defined as >2 SD below the mean volume.

Ratios were calculated according to the following formula: volume ratio = (L - R)/(L + R). A ratio of 0 indicated perfect symmetry. Increasing positive numbers represent relative right atrophy and de- creasing negative numbers represent relative left at- rophy. Surgery

Patients with nonlesional MTLE underwent an- teromedial temporal lobectomy and hippocampec- tomy in a procedure previously described (9). Pathological analysis of resected tissue included ab- solute neuronal cell counts in hippocampal fields cornu ammonis (CA)l-CA4, the dentate, and the subiculum, as well as light microscopy and immu- nohistochemistry. MTS was defined as neuronal loss >50% in CA1 compared with autopsy controls, with evidence of reorganization in the dentate mo- lecular layer. Outcome

Outcome of surgery was assessed at least 12 months after surgery using patient and family re- ports. For this study, complete absence of any sei- zures postoperatively was considered an excellent outcome. Analysis

The presence or absence of hippocampal atro- phy, based on either ratios or absolute numbers, in the context of documented unilaterality or bilater- ality of hippocampal seizure onset, was related to surgical outcome and pathological findings in the resected tissue.

RESULTS MRI

We identified 53 patients who had MRI volumet- ric hippocampal measurements, unilateral or bilat- eral medial temporal lobe seizure onsets recorded by depth electrodes, and anteromedial temporal lobectomy with >1 year of follow-up. Five (9.4%) of these patients had bilateral hippocampal atrophy, defined as absolute hippocampal volume >2 SD be- low the mean volume on both sides.

For the group of five patients, the mean volume of the left hippocampus was 0.765 cc (range, 0.59- 1.04); of the right, it was 0.846 cc (range, 0.58-1.04).

Epilepsia, Vol. 36, No. 9, 1995

BILATERAL HIPPOCAMPAL ATROPHY IN MTLE 907

TABLE 1. Patients with bilateral Hc atrophy

Patient No. of seizures Cell loss

CAllCA3 ~~

no. LHC onset RHC onset L R (%) Surgical outcome

1 0 2 0 3 0 4 4 5 1

4 1.04 0.81 81/79 Seizure-Free 3 0.59 0.58 72/33 Seizure-Free 5 0.94 1.04 621- Seizure-Free 8 0.99 0.84 73/33 Seizure-Free 3 0.82 0.96 81/55 Died postoperatively

Normal hippocampal (Hc) volumes: L, 1.28 ? 0.11 cc: R. 1.28 ? 0.12 cc. All surgeries were done on the right side. LHC, left hippocampus; RHC, right hippocampus.

Only one patient had a significantly (<2 SD) abnor- mal ratio; thus, four of five patients would not have been identified as having hippocampal atrophy us- ing our ratio scores. Results are summarized in Ta- ble 1.

Figure 1 shows a coronal T1-weighted MRI dem- onstrating bilateral hippocampal atrophy, left greater than right.

EEG In two of the five patients with bilateral hippo-

campal atrophy, seizures were recorded indepen- dently from both medial temporal lobes (MTLs), while in the other three, all seizures began in one MTL. However, in the two bilateral-atrophy pa- tients, at least 65% of ictal onsets were recorded from the hippocampus that was eventually re- sected. The other three patients were classified as “unilateral” seizure onset.

Figure 2 shows EEG results from patient 3.

Surgery In all cases, anteromedial temporal lobectomy

with hippocampectomy was performed based on an exclusive or predominant side of ictal onset as de- termined by intracranial recording.

Pathology and outcome Pathological examination revealed hippocampal

sclerosis in all five patients. Three of the four pa- tients with symmetric bilateral hippocampal atro- phy and the one patient with asymmetric bilateral atrophy were seizure-free at latest follow-up (mean, 30 months; range, 12-50). One patient died of status epilepticus in the postoperative period.

DISCUSSION

MRI volumetric demonstration of hippocampal atrophy has been shown to correlate with neuro- pathological evidence of hippocampal cell loss (10,l l), severity of mesial temporal sclerosis (9, and outcome after temporal lobectomy (12). Quan- titative MRI is the noninvasive technique with the highest sensitivity and specificity in detecting MTS,

the most common pathological lesion in patients with MTLE. MRI has a sensitivity of 65% and a specificity of 80% compared with pathological data (3), and of 55 and 78%, respectively, when com- pared with ictal intracranial EEG localization (1 3). Therefore, MRI has been used increasingly to iden- tify patients with MTLE for surgical treatment.

Pathological studies have shown that MTS is present bilaterally to some extent in 72% (14) to 90% (15) of patients. Babb and Brown (16) showed pathological evidence for asymmetric bilateral hip- pocampal damage in 80% of their patients, symmet- ric damage in lo%, and unilateral hippocampal scle- rosis in 10%. Thus, the degree of bilateral involve- ment varies. Our results also show that almost 10% of our patients with MTLE, as determined by intra- cranial EEG, had bilateral hippocampal atrophy that was sufficiently symmetric not to be detected by traditional hippocampal volume ratios. This find- ing suggests that bilateral typical MTS is unusual. However, based on absolute volume criteria, bilat- eral hippocampal disease of asymmetric, but signif-

FIG. 1. Coronal T1-weighted MRI demonstrates bilateral hippocampal atrophy, left greater than right.

Epilepsia, Vol. 36, NO. 9, 1995

D . KING ET AL.

3-5

908

A

I

LPT 1-2 2-3 3-4

- A-

- I A

4-5 LFP 1-3 -

3-5 5-7 - 3-5 --- 5-7 - 3-5 7 5-7 -

LPF 1-3 3-5

RPT 1-2 2-3 3-4 4-5

RFP 1-3 3-5- . ..-. 5-7 A . - -" 3-5- * m__

5-7 ~ - . -..---- 3-5

- - %

,-- rn

LAT 1-3 A

I _J_ - LFT 1-3

I_ __r

5-7 - - -

RAT 1-3 - RFMT 1-3 -

3-5 & s-cs .-- 1-.. ,.. 5-7 2 1 - _ -*.. - .* .,am ZS - *

R

B LPT 1-2 u I

2-3 3-4 4-5

LFP 1-3 _I_czrzlh_ -

3-5 - -- 5-7 -- ?

7

LFT 1-3 3-5 / 5-7 1

3-5 LPF 1-3 /

5-7 % d-/---

RPT 1-2 2-3

Epilepsia, Vol. 36, No. 9, 1995

BILATERAL HIPPOCAMPAL ATROPHY IN MTLE 909 LPT 1-2

2-3 3-4 4-5

3-5 - - 4

5-7 - Y 4 _y - LAT 1-3

3-5

LFT 1-3 3-5 5-7

LPF 1-3

_ _ -

LFP 1-3: - - ,-. _v. I -

5-7 /

3-5 5 5-7 _N__ _c

RPT 1-2 1 2-3 [

_v_

3-4 *

RFMT 1-3

n

LPT 1-2 Nv--cx--v- - v__* - -

~ L-J

3-4 4-5

LFP 1-3 - A -

5-7 LFT 1-3

3-5 /

FIG. 2. Continued. C and D: Evolution of the seizure. The ictal discharge never involves the left medial temporal structures. The patient is seizure-free 2 years after a right anterior temporal lobectomy with hippocampectomy; pathological examination showed hippocampal sclerosis with reorganization. [LPT and RPT, left and right medial temporal depth electrodes; LFP, RFP, LAT, RAT, LFT, RFMT, LPF, and RMT, left and right frontoparietal, anterior temporal, frontotemporal, frontomidtemporal pari- etofrontal, and midtemporal subdural strips. Strips are named for insertion and target area, in that order. Electrode contacts are numbered from 1 (deep) to 7 (superficial). Each division = 1 s; Full scale (FS) = 500 pV.

Eprlepsia, Vol. 36, No. 9, I995

910 D. KING ET AL.

icant, degree was not detected in the others. This discrepancy may reflect the strictness of definitions used to define hippocampal sclerosis in the patho- logical analyses, as well as our requirement of a 2-SD difference for the diagnosis of hippocampal atrophy by MRI. We used only the posterior por- tion of the hippocampus to obtain our absolute vol- umes, but we believe that this represents a true re- flection of the total hippocampal volume based on published literature (17), qualitative studies at our own institution (18), and quantitative reanalysis of the entire hippocampal volume on three patients, which confirmed the representative nature of this measurement.

All patients had excellent surgical outcomes when the temporal lobe of exclusive or predominant ictal onset was resected (although one patient died postoperatively). These observations confirm the findings of Hirsch et al. (6), who showed that the presence of bilateral independent medial temporal seizure onsets is not a contraindication to temporal lobectomy and suggested that the same is true for patients with bilateral hippocampal atrophy.

Our findings contrast, however, with those re- ported by Jack et al. (12) in 50 patients with MTLE studied with scalp EEG and quantitative MRI. They suggested that a good outcome was correlated with the volume of the operated hippocampal formation (HF) and that “the more atrophic the nonoperated HF was, the worse the outcome was.” Jack et al. hypothesized that in patients with bilateral disease of symmetric proportion, “a unilateral ablative pro- cedure would be contraindicated.” One possible explanation for the difference in our results is that we utilized intracranial EEG recordings to docu- ment the predominant side of seizure onset, while Jack et al. relied on scalp recordings to determine the side of ictal onset. The laterality of seizure onset may have been different in their patients had intra- cranial EEG been used. We do not believe that MRI is necessarily sufficient to select the site of resec- tion in these situations, nor can it be a surrogate for invasive EEG studies. There are no other reports of the predictive value of bilateral hippocampal atro- phy for surgical outcome.

Our results confirm that MTS can be a bilateral disease, manifested by bilateral hippocampal atro- phy. Absolute hippocampal volumes, as well as ra- tios, however, are needed to detect some patients with bilateral atrophy, because the atrophy and the

MTS are roughly symmetric in about 10% of pa- tients with MTLE. Although MRI volumetry is a sensitive and specific noninvasive tool for the diag- nosis of mesial temporal sclerosis, it is not sufficient by itself to choose the side for surgery in patients with bilateral hippocampal atrophy.

REFERENCES I .

2.

3.

4.

5 .

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

McCarthy G, Luby M. Structural imaging of the temporal lobe in epilepsy. CIin Neurosci 1994;2:82-8. Spencer S S . Selection of candidates for invasive monitoring. In: Cascino G, Jack CR, eds. Neuroimaging in epilepsy. Stoncham, MA: Butterworth-Heinemann (in press). Spencer SS, McCarthy G, Spencer DD. Diagnosis of medial temporal seizure onset: relative specificity and sensitivity of quantitative MRI. Neurology 1993;43:2117-24. Jack CR Jr, Sharbrough FW, Twomey CK, et al. Temporal lobe seizures: lateralization with MR volume measures of the hippocampal formation. Radiology 1990; 175:423-9. Cascino GD, Jack CR, Parisi JE, et al. Magnetic resonance imaging-based volume studies in temporal lobe epilepsy: pathological correlations. Ann Neurol 1991;30:31-6. Hirsch LJ, Spencer SS, Spencer DD, Williamson PD, Matt- son RH. Temporal lobectomy in patients with bitemporal epilepsy defined by depth electroencephalography. Ann Neurol 1991 ;30:347-56. Spencer SS, Katz A. Amving at the surgical options for intractable seizures. Semin Neurol 1990; 10:422-30. McCarthy G, Riker R, Spencer DD. The stereotactic place- ment of depth electrodes in epilepsy. In: Luders H, ed. Sur- gery of epilepsy. New York: Raven Press, 1991:385-93. Spencer DD, Spencer SS, Mattson RH, Williamson PD, Novelly RA. Access to the posterior medial temporal lobe structures in the surgical treatment of temporal lobe epi- lepsy. Neurosurgery 1984;15:667-71. Lencz T, McCarthy G, Bronen R, et al. Hippocampal in temporal lobe epilepsy: correlation of presurgical MRI vol- umetrics with postsurgical cell counts. Epilepsia 1990;3 1: 667-8. Lencz T, McCarthy G, Bronen RA, et al. Quantitative mag- netic resonance imaging in temporal lobe epilepsy: relation- ship to neuropathology and neuropsychological function. Ann Neurol 1992;3 1:629-37. Jack CR Jr, Sharbrough FW, Cascino GC, Hirschorn KA, O’Brien PC, Marsh WP. Magnetic resonance image-based hippocampal volumetry: correlation with outcome after tem- poral lobectomy. Ann Neurol 1992;31: 138-46. Spencer SS. The relative contributions of MRI, SPECT and PET in epilepsy. Epilepsia 1994;35(suppl 6):S72-89. Margerison JH, Corsellis JAN. Epilepsy and the temporal lobes. Brain 1966;89:499-530. Mouritzen-Dam A. Hippocampal neuronal loss in epilepsy and after experimental seizures. Acra Neurol Scand 1982; 66:601-42. Babb TL, Brown WJ. Pathological findings in epilepsy. In Engel J Jr, ed. Surgical treatment of the epilepsies. New York: Raven Press, 198751 1-39. Kim JH, Tien RD, Felsberg GJ, Osumi AK, Lee N. MR measurement of the hippocampus for lateralization of tem- poral lobe epilepsy: value of measurements of the body vs. the whole structure. AJR 1994;163:1453-7. Bronen R, Fulbright RK, Kim JH, Spencer SS, Spencer DD, Al-Rodhan NRF. Regional distribution of MR findings in hippocampal sclerosis. AJNR 1995;16: 1193-200.

Epilepsia, Vol. 36. No. 9. 1995