Resonance Imaging, Vol. 13, No. 8, pp. 1065-1071, 1995 Copyright 0 1995 Elsevier Science Inc. Printed in the USA. All rights reserved 0730-725x/95 $9.50 + .oo
MRI VOLUMETRICS AND TEMPORAL LOBE SUBSTRATES IN MEDIAL TEMPORAL LOBE EPILEPSY
D. SPENCER,? JUNG H. KIM,? NIHAL DELANEROLLE,? AND GREGORY MCCARTHY *t “Neuropsychology Laboratory, VA Medical Center, West Haven, CT 06516 TDepartment of Surgery (Neurosurgery), Yale University School of Medicine, New Haven, CT 06510, USA MARIE
Forty-nine consecutive patients undergoing anteromedial temporal lobe resection for medically intractable temporal lobe seizures, and averaging 2 yr (range 6 mo to 4 yr) postoperative follow-up, were selected for a retrospective study. This study correlated magnetic resonance imaging (MRI) derived bippocampal volumetrics, preoperative demographics, postoperative seizure control, and tissue analysis, including hippocampal CA (comu ammonis) field neuronal, and glial cell counts, and immunohistochemistry (IHC) evidence for dentate sprouting and reorganization. These measures were compared in hippocampi with or without an adjacent presumptive epileptogenic temporal lobe mass. Mesial temporal sclerosis (MTS) was defined as >SOQo neuronal cell loss averaged across all CA fields with NPY (neuropeptide-y) and somatostatin reorganization. These patients may or may not include granule cell sprouting as determined by dynorphin staining. Patients were divided into two groups based on CA field neuronal cell counts, one averaging >SOQo cell loss and one averaging
tive MRI readings made by trained neuroradiologists.4 Hippocampal atrophy demonstrated by MRI volumetrics in medial temporal lobe epilepsy patients correlates significantly with hippocampal field neuronal loss and glial cell proliferation (hippocampal sclerosis).2,5-7 These clinicopathological correlations suggest that this specific epileptogenic brain region may be identified by static imaging of anatomy. The purpose of this paper is to continue to investi-
MRI volumetry has been an effective adjunct in quantifying the selective destruction of brain tissue by a wide variety of diseases. In particular, volumetrics has proven to be a valuable noninvasive tool in the diagnosis of hippocampal atrophy in patients with classical medial temporal lobe epilepsy. l-3 In addition, it has been shown to be a more accurate measure of disease than qualita-
Haven, CT 06516, USA. E-mail: [email protected]
Address correspondence to Marie Luby, M.Eng., Neuropsychology Laboratory
116B1, VA Medical Center, West 1065
Magnetic Resonance Imaging 0 Volume 13, Number 8, 1995
gate the relationship between MRI volumetrics and microscopic tissue analysis in all patients undergoing medial temporal resection for presumed medial temporal lobe epilepsy with or without a medial temporal lobe mass. MRI hippocampal volumetrics and neuronal cell counts were correlated in this study to determine whether hippocampal volumetrics were sensitive to varying degrees of neuronal loss and the presence of more subtle dentate reorganization seen in the majority of medial temporal lobe epilepsy patients with sclerosis.’ In addition, were volumetrically atrophic hippocampi uniformly the sole epileptogenic region as determined by seizure outcome following selective medial temporal lobe resection? Finally, we address the relationship of volumetrics and tissue analysis in medial temporal lobe epilepsy associated with masses, which by themselves are presumed to be the epileptogenie substrate. The following categories characterize the groups to be analyzed. 1. MTS: Classical mesial temporal sclerosis in patients with medial temporal lobe epilepsy and without masses that exhibit neuronal cell loss, NPY (neuropeptide-y), and somatostatin reorganization.8 These patients may or may not exhibit dentate granule cell collateral sprouting. Three patients in the MTS group did not have >50% cell loss, but they were included in the group based on significant NPY (neuropeptide-y) and somatostatin reorganization. 2. Paradoxical medial temporal lobe epilepsy (paradoxical group): Patients without masses that do not exhibit significant neuronal cell loss, >50%, or any reorganization and were selected for anteromedial temporal resection based on depth electrode ictal recordings. 3. The mass lesion patients are referred to as the mass group and all masses in this group were medial to or included the fusiform gyrus. All of the patients in this study underwent an anteromedial temporal resection that included a hippocampectomy. Hippocampal volumetrics and tissue analysis were performed on all patients as well as assessments of seizure control for an average of 2 yr following surgery. METHODS Subjects All of the patients in this study (N = 49), 29 males, 41 right-handed and 3 ambidextrous, with mean age of 33 yr (range 17-5 l), were selected for an anteromedial temporal resection by the Yale Epilepsy Program (YaleNew Haven Hospital, West Haven VA Hospital) from
1992- 1994. This was based on the concordance of hippocampal atrophy and other noninvasive seizure-localizing criteria such as continuous scalp audiovisual EEG monitoring, cognitive testing, and intracranial amytal procedure.’ If these data were not concordant, the decision to operate was dependent upon invasive audiovisual EEG depth and subdural electrode monitoring. One patient population was categorized as the medial temporal lobe epilepsy (N = 43) receiving either a left (N = 25) or a right (N = 18) medial temporal lobectomy, and the second patient population was categorized as the mass lesion group (N = 6) receiving a left (N = 2) or a right (N = 4) resection that included the temporal lobe pole, the lesion, and a variable amount of the hippocampus. Informed consent was obtained from all subjects before participation in this study. The control group consisted of right-handed professional medical personnel (N= 22), 11 males, from the Veterans Administration Medical Center (West Haven, CT) and Yale University (New Haven, CT) with a mean age of 33 yr (range 23-60). MRI Protocol The MRI protocol performed on the patients (N = 49) consisted of a sagittal Tl (5/skip 2.5 mm), coronal T2 (5/skip 2.5 mm), axial T2 (5/skip 2.5 mm), and coronal SPGR (3 or 5 mm) series that were acquired on a GE Signa 1.5 T clinical magnet (General Electric Inc., Schenectady, NY). The coronal SPGR series (160 mm FOV, 2515 TRITE, 2 nex, 256 x 192 matrix), was used for measurement purposes. An axial SPGR series, with the same parameters as the coronal series except 2 mm thick slices and a FOV range of 200-260 mm, was acquired and resliced coronally for measurement on the control group (N = 22). Measurement Protocol A description of the measurement protocol used in this study can be found in Spencer et al2 and McCarthy and Luby.’ Briefly, the measurement protocol involved the following: Head rotation, if present, was corrected in the original coronal series by choosing the corresponding VIII cranial nerves from each side. The original coronal series or the resliced coronal series, if rotation was present, was then normalized by slicing the data orthogonal to the long axis of the hippocampus. For the control group, oblique coronal slices were created from the original axial series by choosing two symmetrical anatomical landmarks, the internal auditory meati (IAM). These slices were checked for head rotation and then normalized with the same method as the patients’ coronal sequences. The body of the hippocampus was measured bilaterally with a cursor-controlled outlining method from
Medial temporal lobe epilepsy 0 M. LUBY ET AL.
3 mm anterior to the colliculi extending 15 mm in length. This restriction was enforced due to the difficulty in distinguishing the pes hippocampus from the amygdala. Area pixel counts were obtained for each level of hippocampus and were then summed separately and multiplied by slice thickness to produce the left and right hippocampal volumes. Student two-tailed unpaired t-tests were performed to compare each patient group’s hippocampal volumes with the control hippocampal volumes. Left and right hippocampal zscores (Hpzscore) were defined for each patient using Eq. (1).
- ( aigz$$q
Hpzscore = standard deviation for hippocampal volume of control group i 1 (1) Tissue Analysis Tissue analysis, including CA field neuronal and glial cell counts and immunohistochemistry (IHC) evidence for NPY (neuropeptide-y), somatostatin, and dentate granule cell sprouting and reorganization, was performed on all of the patients? Details on how cell counts are performed at this institution can be found in Kim et al.iO and Lencz et a1.5 In this study, cell counts were used from the fields CAl, CA2, CA3, CA4, and granular layer of the dentate gyrus, and compared to the respective cell counts, which were obtained in an autopsy control group population (N= 14), mean age of 38 yr, who all died of nonneurological causes. An average zscore of neuronal cell density, NCD, was calculated for each patient by averaging the separate CA and granular field zscores; zscores were calculated using Eq. (2). Pearson correlation and linear regression analyses were performed between the neuronal cell density (NCD) and the Hpzscore for the resected hippocampus for the medial temporal lobe epilepsy (N = 43) and the mass (N = 6) groups.
Patient CA1 average control cell count ( > ( CA1 cell count > standard deviation of ( control CA1 cell count > (2) RESULTS
Significant volumetric defined as a hippocampal of ~-2.0, indicating that more standard deviations
hippocampal atrophy was zscore, (Hpzscore), value the hippocampus was 2 or smaller in volume than the
control hippocampus with a confidence interval of 95% (p < .05). The average hippocampal volumes, ipsilatera1 and contralateral to the resected medial temporal lobe for each patient group, were compared to the average control hippocampal volume, 1.52 cm3 (Fig. 1). Figure 2A, 2B, and 2C consists of sample patient MRI coronal slices, respectively, from the MTS, paradoxical, and mass groups. Table 1 outlines the percent hippocampal volume loss in the patient groups. In the MTS group (N = 38), 89% exhibited significant ipsilateral volumetric hippocampal atrophy (six bilateral) with a mean volume of 0.92 cm3 (Fig. 2A). The mass group overall had significant hippocampal atrophy when compared to the control hippocampal volume. In the mass group (N = 6), two patients of six exhibited significant ipsilateral volumetric hippocampal atrophy with a mean volume of 1.26 cm3 (Fig. 2B). These two patients had significant bilateral hippocampal atrophy as well, one with a ganglioglioma (19% cell loss) and one with an AVM (14% cell loss). Three mass patients just had significant contralateral volumetric hippocampal atrophy, two with astrocytomas (25.3% and 34% cell losses) and one with a ganglioglioma (36.2% cell loss). The paradoxical medial temporal lobe epilepsy group (N = 5) had one patient with significant hippocampal atrophy, but did not have significant cell loss (24% cell loss). Overall the paradoxical group did not have hippocampal atrophy, with a mean hippocampal volume of 1.46 cm3 (Fig. 2C). MTS was defined as a neuronal cell loss of >50% from the control group that corresponded to a neuronal cell density zscore (NCD) of -2.14 or less. Ninetytwo percent of the MTS (N = 38) patients showed significant cell loss with an average of 60.1% cell loss (Table 1). The three patients who did not have >50% cell loss but were included in the MTS group had significant NPY (neuropeptide-y), somatostatin, and dentate sprouting and reorganization, which distinguished them from the paradoxical medial temporal lobe epilepsy group. None of the paradoxical medial temporal lobe epilepsy patients (N = 5) or the mass patients (N = 6) showed >50% cell loss. Two-tailed unpaired student t-tests revealed that there were significant differences between the mass and medial temporal lobe epilepsy groups in ipsilateral Hpzscore (p = .009) and neuronal cell density (NCD) (p < .OOl). Similarly, there were significant differences between the paradoxical and medial temporal lobe epilepsy groups in Hpzscore (p = .008), neuronal cell density (NCD) (p < .OOl), as well as age of recurrent seizures (p < .OOl). There were no significant differences in Hpzscore, neuronal cell density (NCD), or age of recurrent seizures between the mass and the paradoxical groups. Pearson correlations and linear regressions analyses for the medial temporal lobe epilepsy (MTS and
Magnetic ResonanceImaging 0 Volume 13, Number 8, 1995
Qlpsilateral Hippocampus 0 Contralateral Hippocampus
*- 1.2 E 9 g 1.0 2 5 0.8 E g 0.6 8 i 0.4 0.2 0.0 MASS (N=6)
Fig. 1. The average ipsilateral and contralateral volumes (cm3) with standard errors for the three patient groups plotted against the average control hippocampal volume. The MTS and mass groups demonstrated significant ipsilateral hippocampal atrophy, whereas the paradoxical medial temporal lobe epilepsy group did not.
paradoxical) group (N = 43) were performed comparing the ipsilateral to surgery Hpzscore and neuronal cell density (NCD) (Fig. 3). The medial temporal lobe epilepsy group (N = 43) had a significant correlation between the ipsilateral Hpzscore and neuronal cell density (NCD), r = .40 with p < .008. An insignificant negative correlation between the ipsilateral Hpzscore and neuronal cell density (NCD) was found for the mass group (N = 6), r = -.75 with p < .08. The trend suggested that the larger the hippocampus the greater the cell loss. This may have been caused by the lesion volume being partially included in the hippocampal measures in some patients when the two regions were anatomically indistinguishable on MRI, as displayed in Fig. 2C. Patients were evaluated postoperatively for seizure reduction and categorized as successesif their seizures were completely stopped with or without the occurrence of auras. The successrate was highest in the MTS group
(N= 38), 76% were seizure free, and data was not available for two patients (Table 1). Four of the six failures in the MTS group experienced some reduction, approximately 25070,in seizure frequency. The mass group had a 67% successrate, one patient experienced an increase of 25% in seizures and one experienced a decrease of 25% in seizures. The paradoxical group had a 60% success rate, one patient receiving a 25% reduction in seizure frequency. DISCUSSION Significant hippocampal atrophy identified by presurgical MRI volumetrics was predictive of significant neuronal cell loss in 91% of the patients in this study who had medial temporal lobe epilepsy. This significant correlation between neuronal cell counts and hippocampal volumetrics has been previously reported.5*6 In addition, the MTS patients had the highest success rate, 76%) when compared to the paradoxical mesial
Table 1. Patient statistics by group % Febrile Patient group MTS (N = 38) Paradoxical (N = 5) Mass (N = 6)
Age of recurrent seizures (yr) 10.09 (6.62)
15.33 (13.60) 21.00 (1.00)
% Hippocampal volume loss
070Neuronal cell loss
39.75 (13.12) 3.76 (15.94)
17.78 (5.55) 26.12 (8.63)
VoSuccess rate 76 60 67
Medial temporal lobe epilepsy 0 M.
(4 Fig. 2. (A) Patient SG is a 3%yr-old ambidextrous male who underwent a left AMTR. Preoperatively, MRI volumetrics showed significant left hippocampal atrophy, LH = 1.02 cm3 and RH = 1.50 cm3 with left Hpzscore = -2.89. After resection, tissue analysis showed 66.7% overall cell loss, which was significant, NCD = -3.65. He has been seizure free for the last 4 mo. (B) Patient PC is a 37-yr-old right-handed female who underwent a right anteromedial temporal lobe resection. Preoperatively, MRI volumetrics showed no hippocampal atrophy, LH = 1.54 cm’, and RH = 1.49 cm3 with right Hpzscore = -0.16. After resection, tissue analysis showed 10% overall cell loss that was not significant, NCD = -0.67. She has been seizure free for the last 2yr since her surgery. (C) Patient RC is a 31-yr-old right-handed female who underwent a left temporal lobe mass resection with a partial hippocampal resection. Preoperatively, MRI showed a mass lesion in her left temporal lobe that included the hippocampus. Volumetrics showed no hippocampal atrophy, LH = 1.43 cm3 and RH = 1.24 cm3 with left Hpzscore = -0.60. After resection, tissue analysis showed a 34% overall cell loss that was not significant, NCD = - 1.82, and pathology classified the mass lesion as an astrocytoma.
temporal lobe (60%) and the mass lesion patients (67%). This success rate is consistent with those reported by other studies.2,3,6 However, hippocampal atrophy as shown by MRI volumetrics was not indicative of significant cell loss in all of the mass lesion patients that demonstrated significant volumetric hippocampal atrophy. None of the mass lesion patients had cell loss on the scale of the classical MTS patients, but two out of six had comparable
bilateral hippocampal atrophy as shown by volumetrics. In addition, significant volumetric hippocampal atrophy was not necessary for a successful outcome for the mass patients. This is not entirely consistent with the study by Cascino et al.,” in which three patients subsequently received temporal lobcctomies with amygdalohippocampectomies after resection of a temporal lobe lesion, that did not render them seizure free. Preoperative MRI hippocampal volumetrics revealed that
Magnetic Resonance Imaging l Volume 13, Number 8, 1995
I -4.5 I?
A A .
d : .
. . ,.a.
-- -6 .
Fig. 3. The distribution of the quantities, ipsilateral Hpzscore, and neuronal cell density zscore (NCD), is plotted for the three patient groups to demonstrate the differences in hippocampal atrophy and cell loss between the paradoxical and mass medial temporal lobe epilepsy groups when compared to the MTS group.
all three of these patients had hippocampal atrophy ipsilateral to the surgical side.” However; in only one patient was tissue available to prove hippocampal sclerosis by pathology in that study. The present study would seem to indicate some ambiguity regarding whether or not the degree of cell loss seen by patients with a mesial temporal lobe mass is detected accurately by MRI volumetrics, and if volumetrics is reliable on its own to detect dual pathology patients and to dictate resection of the hippocampus along with the mass. It has been proposed that the pathology of the mass may help one decide whether or not the hippocampus should be resected when the mass does not involve the hippocampus. In a study by Levesque et al.,” the presence of hamartomas was correlated with severe cell loss, >30%, whereas gliomas were associated with mild cell loss, ~30%. Using the >30% cell loss as the criterion for significant
cell loss, two patients with masses
in this study, both with gangliogliomas, had >30% cell loss, one with 34% cell loss and normal volumetrics and one with 36% cell loss and significant contralateral volumetric hippocampal atrophy. MRI volumetrics failed to show significant atrophy of the ipsilateral hippocampus in either of these patients. Finally, the paradoxical medial temporal lobe epilepsy group of patients remain an enigma. Four out of five of these patients’ hippocampi appear anatomically normal by MRI volumetrics and all appear normal by postoperative cell counts. Although seizure control is poorer than the MTS or the mass groups, still 60% of
the paradoxical group are controlled by resection following invasive electrophysiological localization. One explanation for this is that the substrate was present in the resected neocortex or amygdala and overlooked by our analysis. Alternatively, the epileptogenic region may be broad and outside the hippocampus but may be rendered quiescent by disconnecting the critical medial temporal network. We have also pointed out that the hippocampus may be atrophic bilaterally in MTS and contralateral to a medial temporal lobe mass and yet seizures are controlled if an atrophic hippocampus is left. These data stress that identification of appropriate anatomical substrates usually thought to be epileptogenic is increasingly possible with MRI, but must be concordant with electrophysiological abnormalities before one can reasonably be certain of localization. REFERENCES McCarthy, G.; Luby, M. Imaging the structural changes associated with human epilepsy. Clin. Neurosci. 2:8288; 1994. Spencer, S.S.; McCarthy, G.; Spencer, D.D. Diagnosis of medial temporal lobe seizure onset: Relative specificity and sensitivity of quantitative MRI. Neurology 43: 2117-2124; 1993. Kuzniecky, R.I.; Cascino, G.D.; Pam&A.; Jack, C.R., Jr.; Berkovic, S.F.; Jackson, G.D.; McCarthy, G. Structural neuroimaging. In: J. Engel, Jr. (Ed). Surgical Treatment of the Epilepsies, 2nd ed. New York: Raven Press; 1993:pp. 197-209.
Medial temporal lobe epilepsy 0 M.
4. Cendes,F.; Leproux, F.; Melanson, D.; Ethier, R.; Evans,A.; Peters,T.; Andermann, F. MRI of amygdala and hippocampusin temporallobeepilepsy.J. Comput. Assist. Tomogr. 17(2):206-210;1993. 5. Lencz, T.; McCarthy, G.; Bronen, R.A.; Scott, T-M.; Inserni, J.A.; Sass,K.J.; Novelly, R.A.; Kim, J.H.; Spencer,D.D. Quantitative magneticresonanceimaging in temporallobe epilepsy:Relationshipto neuropathology andneuropsychologicalfunction. Ann. Neurol. 31: 629-637; 1992. 6. Cascino,G.D.; Jack, C.R., Jr.; Parisi,J.E.; Sharbrough, F.W.; Hirschorn, K.A.; Meyer, F.B.; Marsh, W.R.; O’Brien, P.C. Magneticresonance imaging-based volume studiesin temporal lobe epilepsy: Pathologic correlations. Ann. Neurol. 30:31-36; 1991. 7. Kuzniecky, R.; de la Sayette, V.; Ethier, R.; Melanson, D.; Andermann,F.; Berkovic, S.; Robitaille,Y.; Olivier, A.; Peters,T.; Feindel,W. Magnetic resonanceimaging in temporal lobe epilepsy: Pathological correlations. Ann. Neurol. 22:341-347; 1987.
8. deLanerolle,N.C.; Kim, J.H.; Brines,M.L. Cellularand molecularalterationsin partial epilepsy.Clin. Neurosci. 2:64-81; 1994. 9. deLanerolle,N.C.; Kim, J.H.; Robbins,R.J.; Spencer, D.D. Hippocampalinter-neuronlossandplasticity in human temporal lobe epilepsy.Brain Res. 495:387-395; 1989. 10. Kim, J.H.; Guimaraes,P.O.; Shen, M.Y.; Masukawa, L.M.; Spencer,D.D. Hippocampalneuronaldensity in temporal lobe epilepsywith and without gliomas.Acta Neuropathol. 80:41-45; 1990. 11. Cascino,G.D.; Jack, CR., Jr.; Parisi,J.E.; Sharbrough, F.W.; Schreiber,C.P.; Kelly, P.J.; Trenerry, M.R. Operative strategyin patientswith MRI-identified dual pathology and temporal lobe epilepsy. EpilepsyRes, 14: 175-182,1993. 12. Levesque,M.F.; Nakasato,N.; Vinters, V.; Babb, T.L. Surgicaltreatmentof limbic epilepsyassociatedwith extrahippocampallesions:The problemof dual pathology. J. Neurosurg. 75:364-370;1991.