- Email: [email protected]
Correlation of seizure frequency with N-acetyl-aspartate levels determined by 1H magnetic resonance spectroscopic imaging
Magnetic
PI1 SO730-725X( 96) 00327-8
ELSEVIER
l
Resonance Imaging, Vol. 15, No. 4, pp. 475-478, 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. All rights reserved 0730-725x/97 $17.00 + .oo
Original Contribution CORRELATION OF SEIZURE FREQUENCY WITH N-ACETYLASPARTATE ‘LEVELS DETERMINED BY ‘H MAGNETIC RESONANCE SPECTROSCOPIC IMAGING P.A. GARCIA,*
K.D. LAXER,* G.B. MATSONS
J. VAN DER GROND,~ AND M.W. WEINER/~
J.W. HUGG,~
*Department of Neurology, University of California, San Francisco, California, USA, TDepartment of Radiology, Academic Hospital Utrecht, $Department of Neurology, University of Alabama, Birmingham, Alabama, USA, Department of Pharmaceutical Chemistry, University of California, San Francisco, California, USA, Department of Medicine, University of California, San Francisco, California, USA Using ‘H MRSI, we measured N-acetyl-aspartate (NAA), a neuronal marker, in the seizure focus of 16 patients with partial epilepsy. Decreasing NAA correlated with increasing seizure frequency in frontal lobe epilepsy (t = -0.72, p < 0.02) and a similar trend was present in temporal lobe epilepsy (r = -.60, p < 0.06). NAA was not related to the duration of epilepsy. We conclude that patients with higher seizure frequency have evidence of greater neuron loss or dysfunction. 0 1997 Elsevier Science Inc. Keywords:
MRSI, magnetic resonance spectroscopic
imaging; Seizure frequency; Epilepsy; Spectroscopy.
METHODS
INTRODUCTION
Sixteen patients with medically refractory partial epilepsy [eight with frontal lobe epilepsy (FLE) and eight with temporal lobe epilepsy (TLE)] consented to be studied (Table 1) . Patients with expanding mass lesions on MRI were excluded. One patient with FLE had a cavernous angioma. Two TLE patients showing hippocampal T2 signal prolongation without mass effect had small gangliogliomas discovered on pathological examination. Seizure localization was determined by ictal electroencephalographic (EEG) recordings which were sufficiently localizing for the patient to be considered a candidate for surgical resection. All but
Many patients with partial epilepsy experience recurrent seizures despite treatment with anticonvulsants. In these patients, seizure frequency and duration of epilepsy are measures of clinical severity. Seizure frequency correlates with the probability of seizure remission’ as well as the perception of well-being.’ Unfortunately, we do not understand how the clinical severity is related to the pathological substrate. Proton magnetic resonance spectroscopic imaging (‘H MRSI) permits non-invasive measurement of Nacetyl-aspartate (NAA), a neuronal marker. ‘H MRSI consistently demonstrates decreased NAA in conditions characterized by neuronal loss such as partial epilepsy,3 stroke4 and glial tumors.5 In animal models of status epilepticus, NAA measurements correlate with pathologically determined neuron 10~s.~ We designed this study to determine whether NAA is related to seizure frequency or epilepsy duration in patients with partial epilepsy.
two of the patients with FLE required intracranial
re-
cordings to fulfill the criteria while the TLE patients had adequate localization
with scalp EEG recordings.
NAA in all of these patients had previously been found to be decreased in the epileptogenic focus compared to the contralateral homologous region.3,7 MRI and MRSI studies were performed using a Gyroscan S15 whole body imaging and spectroscopy Avenue, University of California, San Francisco, Box 0138, San Francisco, CA 94143-0138, email address: garfish@ itsa.ucsf.edu
311196:ACCEPTED 91496. Address correspondence to Paul A. Garcia, MD, Northern California Comprehensive Epilepsy Center, 400 Parnassus RECEIVED
415
Magnetic ResonanceImaging 0 Volume 15, Number 4, 1997
476
Table 1. Demographic information Patient #I #2 #3 #4 #5 #6 #7 #8 #9 #lO #ll #12 #13 #14 #15 #16
Age (ye@ 29 32 27 32 23 18 13 21 37 20 22 19 41 34 36 39
EEG
MRI
Duration k--s)
Pathology
Surgical outcome
R. Frontal L. Frontal* L. Frontal* L. Frontal L. Frontal L. Frontal L. Frontal L. Frontal L. Temporal* R. Temporal* R. Temporal* L. Temporal* R. Temporal* R. Temporal* L. Temporal* R. Temporal*
NL NL AVM NL NL NL NL NL IHS HA IHS HA NL HA HA HA
22 5 11 10 12 7 3 17 7 7 21 8 3 18 34 12
gliosis ND AVM gliosis chronic inflammation cortical dysplasia gliosis ND ganglioglioma MTS ganglioglioma MTS MTS ND MTS MTS
>90% decrease ND 100% decrease ~50% decrease >90% decrease No improvement >90% decrease ND >90% decrease 100% decrease 100% decrease 100% decrease 100% decrease ND 100% decrease 100% decrease
ND = Not done,NL = normal. IHS = Increased hippocampalsignal, HA = hippocampalatrophy. * Scalp recordings only.
system (Philips Medical Systems, Shelton, CT) operated at 2 Tesla. A saddle-type imaging headcoil was used. A vacuum-molded head holder (Vat-Pat, Olympic Medical, Seattle, WA) decreased head movement. Sixteen sagittal localizer (7.1 mm thick, 0.7 mm gap, TR = 500 ms, TE = 30 ms) and 16 axial (7.1 mm thick, 1.4 mm gap, TR = 2500 ms, TE = 30 and 80 ms) slices were obtained. Angulation was -30 degrees from the orbital-auditory meatal plane in all but the first two patients with FLE in whom +30 degree angulation was employed. The magnetic field was then shimmed for a non-localized water resonance linewidth of less than 15 Hz.337 Typical volumes of interest (VOI) excluded the skull measuring 10 X 10 X 5 cm3. VOI measurements were 10 x 10 X 2 cm3 in the first two FLE patients in whom 2-D studies were performed. The double spin-echo (PRESS) sequence used for VOI localization employed amplitude-modulated RF pulses in combination with slice selection gradients. Gradient spoiler pulses in conjunction with spin-echo pulses reduced spurious transverse magnetization. The echo was sampled with a one kHz bandwidth after the last gradient pulse with 21% of the 5 12 ms sampling interval occurring before the echo center at TE = 272 ms. The water signal from the second echo of the PRESS sequence was maximized and gradient spoiler pulses were adjusted to maximize the water echo signal. Localized shimming was performed typically resulting in a water resonance linewidth of less than 6 Hz (0.07 ppm) . Two WEFT water suppression pulses (frequencyselective adiabatic inversion pulses with 50 Hz bandwidth
centered on the water resonance) minimized longitudinal water magnetization and gradient spoiler pulses dispersed spurious transverse magnetization excited by longitudinal inversion. To further suppress skull lipid signal, three cosine-sine outer-volume-suppression pulses were employed between the two WEFT pulses.3.7 Proton spectra were acquired using 3-D gradient phase encoding for all except the first two patients with FLE. An ellipsoid in k-space with 16 X 16 X 12 gradient phase encoding steps along principle axes was typically sampled over a 16 x 16 X 14.1 cm field of view. The phase encoding steps were applied between the 90 degree excitation pulse and the first 180 degree refocusing pulse. A single acquisition was obtained for each step. Spectra from the first two patients with FLE were obtained using 2-D phase encoding with 16 x 16 gradient phase encoding steps over a field of view of 18 X 18 cm.3z7 MRS images were reconstructed using Fourier transforms and metabolite images were generated by spectral integration. Nominal voxel size was 0.85 cm3 for the 3-D studies and 2.5 cm3 for two FLE patients undergoing 2-D studies. MRS images were registered to the corresponding MR image to allow anatomical analysis of spectra. In patients with TLE, spectra were selected from a single voxel centered in the epileptogenie hippocampus; in patients with FLE, spectra were summed from the epileptogenic frontal lobe. Since the volume of interest often contained only a portion of the frontal lobes, summed voxels were selected for spectral analysis from all available frontal lobe tissue.
Seizure
frequency
with N-acetyl-aspartate
Spectra were fit by a least squares method (NMRl program, New Methods Research, Syracuse, NY). Spectra were interpolated by zero-filling and Fourier transformation to a digital resolution of 0.5 Hz/point and the creatine (Cr) and NAA peaks were identified by their chemical shifts.3,7 Complex partial seizures (or simple partial seizures with prominent motor manifestations) were recorded by the patients and/or family members. No patients had frequent (more than one/month) seizures with secondary generalization. Seizure frequency (seizures/ month) was determined at the last pre-imaging clinic appointment (2-8 weeks before the MRSI study). Together with the patients and family members, one of the investigators (KDL) reviewed seizure calendars for the most recent three months to calculate seizure frequency. The ability of the patient to recognize seizures was confirmed during inpatient video/EEG monitoring. The investigators reviewed the seizure history to determine the duration of the seizure disorder i.e. the time from the first afebrile seizure to the time of the MRSI study (Previously reported durations included febrile seizure onset) .3 NAA/Cr ratios were plotted against seizure frequency and duration of epilepsy. Pearson’s correlation coefficient was determined. Since NAA has been found to be decreased in all partial epilepsies studied to date, 3.7,8we expected that NAA would decrease with increasing seizure frequency and duration. Thus, we used one tailed t-tests to determine whether the slope differed from zero. Results were considered significant for p values <0.05.
RESULTS In patients with FLE (Fig. 1) , NAA/Cr decreased as seizure frequency increased (correlation coefficient = -0.72, p < 0.02). In patients with TLE, a trend toward decreased NAA/Cr was seen with increasing seizure frequency (correlation coefficient = -0.60, p < 0.06). In both groups of patients duration of epilepsy correlated poorly with the NAA/Cr ratio (correlation coefficient = 0. IO, p < 0.41 for FLE and correlation coefficient = 0.18, p < 0.34 for TLE).
DISCUSSION Our finding that NAA/Cr decreases as frontal lobe seizure frequency increases suggests that the clinical manifestations of FLE correspond to underlying pathology. The strong trend in patients with TLE suggests a similar relationship in this group. Thus, for the first time, a non-invasive test provides evidence for a direct relationship between clinical and pathophysiological findings in patients with partial epilepsy.
levels 0 P. A. GARCIA ETAL.
1
1.2
I .4
1.6
411
1.8
2
2.2
NAMCr
Fig. 1. NAA vs. seizure frequency. NAA/Cr ratio in patients with TLE (triangles) and FLE (circles) is plotted vs. seizure freuency (FLE on Y, axis, TLE on YZ axis). Regression lines (solid = FLE, broken = TLE) show increasing seizure frequency with decreasing NAA/Cr.
The lack of correlation of epilepsy duration with NAA/Cr ratio likely reflects the fact that epilepsy duration is a poor measure of epilepsy severity. Seizure frequency may have been quite variable for patients during the many years of their epilepsy. Perhaps a lifetime seizure total would be a better measure. Unfortunately, we find that we are often unable to obtain accurate estimates of seizure activity from prior years. Cendes and colleagues did not find a relationship between seizure frequency and NAA in patients with TLE.* In contrast to our methods, they estimated seizure frequency over the prior five years. While this should theoretically be a useful measure, we have found that many patients have difficulty estimating seizure frequency beyond the past few months. Estimates limited to recent times may be more accurate. The less compelling relationship in TLE patients may reflect underlying differences in pathology. Patients with TLE had mesial temporal sclerosis (MTS > or small gangliogliomas, while the patients with FLE had variable pathological findings including vascular malformations, cortical dysplasia, gliosis and chronic inflammation. Alternatively, methodology may explain the clearer relationship in FLE. Frontal measurements included spectra from all available frontal lobe tissue while temporal lobe measurements were limited to a single voxel in the hippocampus. Prior studies have shown widespread neuron loss associated with partial epilepsy.‘,* Measurement of a larger volume of tissue may produce more robust results. From our findings we can not infer a cause-effect relationship between seizure frequency and NAA levels. Nevertheless, this study provides the first step toward understanding this relationship. Seizures may cause neuron loss or dysfunction resulting in decreased
Magnetic ResonanceImaging0 Volume 15, Number 4, 1997
418
NAA. Alternatively, patients with more extensive underlying neuron loss or dysfunction may be susceptible to having more seizures. Serial NAA studies controlled for seizure frequency should enable us to address the issue of causality. are gratefulto JohnWalker,PhD and WilliamJ. Marks,Jr., M.D. for theirhelpfulsuggestions. Dr. Laxeris supported by NIH grantROl-NS31966.
Acknowledgements-We
REFERENCES 1. Callaghan,N.; Garrett, A.; Goggin, T. Withdrawal of anticonvulsantdrugsin patientsfree of seizuresfor two years.N. Engl. J. Med. 318:942-946; 1988. 2. Dodson,W.E. Quality of Life Measurementsin Children with Epilepsy. In: M.R. Trimble and W.E. Dodson, (Eds) . Epilepsy and Quality of Life. New York: Raven Press,Ltd.; 1994: 219. 3. Hugg, J.W.; Laxer, K.D.; Matson, G.B.; Maudsley, A.A.; Weiner, M.W. Neuron loss localizes human focal epilepsy by in vivo proton MR spectroscopicimaging.Ann. Neurol. 34:788-794; 1993.
4. Hugg, J.W.; Duijn, J.H.; Matson, G.B.; Maudsley, A.A.; Tsuruda,J.S.; Gelinas,D.F.; Weiner,M.W. Elevatedlactate and alkalosisin chronic humanbrain infarction observed by 1H and 31P MR spectroscopicimaging. J. Cereb.Blood Flow Metab. 12:734-744; 1992. 5. Arnold, D.L.; Shoubridge,E.A.; Villemure, J.G.; Feindel, W. Proton andphosphorusmagneticresonancespectroscopy of humanastrocytomasin vivo. Preliminary observations on tumor grading. NMR Biomed. 3:184- 189; 1990. 6. Ebisu, T.; Rooney, W.D.; Graham,S.H.; Weiner, M.W.; Maudsley, A.A. N-acetyl-aspartateas an in vivo marker of neuronalviability in kainate-inducedstatusepilepticus: ‘H magneticresonancespectroscopicimaging. J. Cereb. Blood Flow Metab. 14:373-382; 1994. 7. Garcia,P.A.; Laxer, K.D.; van der Grond, J.; Hugg,J.W.; Mattson, G.B.; Weiner, M.W. Proton magneticresonance spectroscopicimaging in patientswith frontal lobe epilepsy. Ann. Neurol. 37:279-281; 1995. 8. Cendes,F.; Andermann,F.; Preul, MC.; Arnold, D.L. Lateralizationof temporallobe epilepsybasedon regional metabolic abnormalitiesin proton magnetic resonance spectroscopicimages.Ann. Neurol. 35:211-216; 1994.
PI1 SO730-725X( 96) 00327-8
ELSEVIER
l
Resonance Imaging, Vol. 15, No. 4, pp. 475-478, 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. All rights reserved 0730-725x/97 $17.00 + .oo
Original Contribution CORRELATION OF SEIZURE FREQUENCY WITH N-ACETYLASPARTATE ‘LEVELS DETERMINED BY ‘H MAGNETIC RESONANCE SPECTROSCOPIC IMAGING P.A. GARCIA,*
K.D. LAXER,* G.B. MATSONS
J. VAN DER GROND,~ AND M.W. WEINER/~
J.W. HUGG,~
*Department of Neurology, University of California, San Francisco, California, USA, TDepartment of Radiology, Academic Hospital Utrecht, $Department of Neurology, University of Alabama, Birmingham, Alabama, USA, Department of Pharmaceutical Chemistry, University of California, San Francisco, California, USA, Department of Medicine, University of California, San Francisco, California, USA Using ‘H MRSI, we measured N-acetyl-aspartate (NAA), a neuronal marker, in the seizure focus of 16 patients with partial epilepsy. Decreasing NAA correlated with increasing seizure frequency in frontal lobe epilepsy (t = -0.72, p < 0.02) and a similar trend was present in temporal lobe epilepsy (r = -.60, p < 0.06). NAA was not related to the duration of epilepsy. We conclude that patients with higher seizure frequency have evidence of greater neuron loss or dysfunction. 0 1997 Elsevier Science Inc. Keywords:
MRSI, magnetic resonance spectroscopic
imaging; Seizure frequency; Epilepsy; Spectroscopy.
METHODS
INTRODUCTION
Sixteen patients with medically refractory partial epilepsy [eight with frontal lobe epilepsy (FLE) and eight with temporal lobe epilepsy (TLE)] consented to be studied (Table 1) . Patients with expanding mass lesions on MRI were excluded. One patient with FLE had a cavernous angioma. Two TLE patients showing hippocampal T2 signal prolongation without mass effect had small gangliogliomas discovered on pathological examination. Seizure localization was determined by ictal electroencephalographic (EEG) recordings which were sufficiently localizing for the patient to be considered a candidate for surgical resection. All but
Many patients with partial epilepsy experience recurrent seizures despite treatment with anticonvulsants. In these patients, seizure frequency and duration of epilepsy are measures of clinical severity. Seizure frequency correlates with the probability of seizure remission’ as well as the perception of well-being.’ Unfortunately, we do not understand how the clinical severity is related to the pathological substrate. Proton magnetic resonance spectroscopic imaging (‘H MRSI) permits non-invasive measurement of Nacetyl-aspartate (NAA), a neuronal marker. ‘H MRSI consistently demonstrates decreased NAA in conditions characterized by neuronal loss such as partial epilepsy,3 stroke4 and glial tumors.5 In animal models of status epilepticus, NAA measurements correlate with pathologically determined neuron 10~s.~ We designed this study to determine whether NAA is related to seizure frequency or epilepsy duration in patients with partial epilepsy.
two of the patients with FLE required intracranial
re-
cordings to fulfill the criteria while the TLE patients had adequate localization
with scalp EEG recordings.
NAA in all of these patients had previously been found to be decreased in the epileptogenic focus compared to the contralateral homologous region.3,7 MRI and MRSI studies were performed using a Gyroscan S15 whole body imaging and spectroscopy Avenue, University of California, San Francisco, Box 0138, San Francisco, CA 94143-0138, email address: garfish@ itsa.ucsf.edu
311196:ACCEPTED 91496. Address correspondence to Paul A. Garcia, MD, Northern California Comprehensive Epilepsy Center, 400 Parnassus RECEIVED
415
Magnetic ResonanceImaging 0 Volume 15, Number 4, 1997
476
Table 1. Demographic information Patient #I #2 #3 #4 #5 #6 #7 #8 #9 #lO #ll #12 #13 #14 #15 #16
Age (ye@ 29 32 27 32 23 18 13 21 37 20 22 19 41 34 36 39
EEG
MRI
Duration k--s)
Pathology
Surgical outcome
R. Frontal L. Frontal* L. Frontal* L. Frontal L. Frontal L. Frontal L. Frontal L. Frontal L. Temporal* R. Temporal* R. Temporal* L. Temporal* R. Temporal* R. Temporal* L. Temporal* R. Temporal*
NL NL AVM NL NL NL NL NL IHS HA IHS HA NL HA HA HA
22 5 11 10 12 7 3 17 7 7 21 8 3 18 34 12
gliosis ND AVM gliosis chronic inflammation cortical dysplasia gliosis ND ganglioglioma MTS ganglioglioma MTS MTS ND MTS MTS
>90% decrease ND 100% decrease ~50% decrease >90% decrease No improvement >90% decrease ND >90% decrease 100% decrease 100% decrease 100% decrease 100% decrease ND 100% decrease 100% decrease
ND = Not done,NL = normal. IHS = Increased hippocampalsignal, HA = hippocampalatrophy. * Scalp recordings only.
system (Philips Medical Systems, Shelton, CT) operated at 2 Tesla. A saddle-type imaging headcoil was used. A vacuum-molded head holder (Vat-Pat, Olympic Medical, Seattle, WA) decreased head movement. Sixteen sagittal localizer (7.1 mm thick, 0.7 mm gap, TR = 500 ms, TE = 30 ms) and 16 axial (7.1 mm thick, 1.4 mm gap, TR = 2500 ms, TE = 30 and 80 ms) slices were obtained. Angulation was -30 degrees from the orbital-auditory meatal plane in all but the first two patients with FLE in whom +30 degree angulation was employed. The magnetic field was then shimmed for a non-localized water resonance linewidth of less than 15 Hz.337 Typical volumes of interest (VOI) excluded the skull measuring 10 X 10 X 5 cm3. VOI measurements were 10 x 10 X 2 cm3 in the first two FLE patients in whom 2-D studies were performed. The double spin-echo (PRESS) sequence used for VOI localization employed amplitude-modulated RF pulses in combination with slice selection gradients. Gradient spoiler pulses in conjunction with spin-echo pulses reduced spurious transverse magnetization. The echo was sampled with a one kHz bandwidth after the last gradient pulse with 21% of the 5 12 ms sampling interval occurring before the echo center at TE = 272 ms. The water signal from the second echo of the PRESS sequence was maximized and gradient spoiler pulses were adjusted to maximize the water echo signal. Localized shimming was performed typically resulting in a water resonance linewidth of less than 6 Hz (0.07 ppm) . Two WEFT water suppression pulses (frequencyselective adiabatic inversion pulses with 50 Hz bandwidth
centered on the water resonance) minimized longitudinal water magnetization and gradient spoiler pulses dispersed spurious transverse magnetization excited by longitudinal inversion. To further suppress skull lipid signal, three cosine-sine outer-volume-suppression pulses were employed between the two WEFT pulses.3.7 Proton spectra were acquired using 3-D gradient phase encoding for all except the first two patients with FLE. An ellipsoid in k-space with 16 X 16 X 12 gradient phase encoding steps along principle axes was typically sampled over a 16 x 16 X 14.1 cm field of view. The phase encoding steps were applied between the 90 degree excitation pulse and the first 180 degree refocusing pulse. A single acquisition was obtained for each step. Spectra from the first two patients with FLE were obtained using 2-D phase encoding with 16 x 16 gradient phase encoding steps over a field of view of 18 X 18 cm.3z7 MRS images were reconstructed using Fourier transforms and metabolite images were generated by spectral integration. Nominal voxel size was 0.85 cm3 for the 3-D studies and 2.5 cm3 for two FLE patients undergoing 2-D studies. MRS images were registered to the corresponding MR image to allow anatomical analysis of spectra. In patients with TLE, spectra were selected from a single voxel centered in the epileptogenie hippocampus; in patients with FLE, spectra were summed from the epileptogenic frontal lobe. Since the volume of interest often contained only a portion of the frontal lobes, summed voxels were selected for spectral analysis from all available frontal lobe tissue.
Seizure
frequency
with N-acetyl-aspartate
Spectra were fit by a least squares method (NMRl program, New Methods Research, Syracuse, NY). Spectra were interpolated by zero-filling and Fourier transformation to a digital resolution of 0.5 Hz/point and the creatine (Cr) and NAA peaks were identified by their chemical shifts.3,7 Complex partial seizures (or simple partial seizures with prominent motor manifestations) were recorded by the patients and/or family members. No patients had frequent (more than one/month) seizures with secondary generalization. Seizure frequency (seizures/ month) was determined at the last pre-imaging clinic appointment (2-8 weeks before the MRSI study). Together with the patients and family members, one of the investigators (KDL) reviewed seizure calendars for the most recent three months to calculate seizure frequency. The ability of the patient to recognize seizures was confirmed during inpatient video/EEG monitoring. The investigators reviewed the seizure history to determine the duration of the seizure disorder i.e. the time from the first afebrile seizure to the time of the MRSI study (Previously reported durations included febrile seizure onset) .3 NAA/Cr ratios were plotted against seizure frequency and duration of epilepsy. Pearson’s correlation coefficient was determined. Since NAA has been found to be decreased in all partial epilepsies studied to date, 3.7,8we expected that NAA would decrease with increasing seizure frequency and duration. Thus, we used one tailed t-tests to determine whether the slope differed from zero. Results were considered significant for p values <0.05.
RESULTS In patients with FLE (Fig. 1) , NAA/Cr decreased as seizure frequency increased (correlation coefficient = -0.72, p < 0.02). In patients with TLE, a trend toward decreased NAA/Cr was seen with increasing seizure frequency (correlation coefficient = -0.60, p < 0.06). In both groups of patients duration of epilepsy correlated poorly with the NAA/Cr ratio (correlation coefficient = 0. IO, p < 0.41 for FLE and correlation coefficient = 0.18, p < 0.34 for TLE).
DISCUSSION Our finding that NAA/Cr decreases as frontal lobe seizure frequency increases suggests that the clinical manifestations of FLE correspond to underlying pathology. The strong trend in patients with TLE suggests a similar relationship in this group. Thus, for the first time, a non-invasive test provides evidence for a direct relationship between clinical and pathophysiological findings in patients with partial epilepsy.
levels 0 P. A. GARCIA ETAL.
1
1.2
I .4
1.6
411
1.8
2
2.2
NAMCr
Fig. 1. NAA vs. seizure frequency. NAA/Cr ratio in patients with TLE (triangles) and FLE (circles) is plotted vs. seizure freuency (FLE on Y, axis, TLE on YZ axis). Regression lines (solid = FLE, broken = TLE) show increasing seizure frequency with decreasing NAA/Cr.
The lack of correlation of epilepsy duration with NAA/Cr ratio likely reflects the fact that epilepsy duration is a poor measure of epilepsy severity. Seizure frequency may have been quite variable for patients during the many years of their epilepsy. Perhaps a lifetime seizure total would be a better measure. Unfortunately, we find that we are often unable to obtain accurate estimates of seizure activity from prior years. Cendes and colleagues did not find a relationship between seizure frequency and NAA in patients with TLE.* In contrast to our methods, they estimated seizure frequency over the prior five years. While this should theoretically be a useful measure, we have found that many patients have difficulty estimating seizure frequency beyond the past few months. Estimates limited to recent times may be more accurate. The less compelling relationship in TLE patients may reflect underlying differences in pathology. Patients with TLE had mesial temporal sclerosis (MTS > or small gangliogliomas, while the patients with FLE had variable pathological findings including vascular malformations, cortical dysplasia, gliosis and chronic inflammation. Alternatively, methodology may explain the clearer relationship in FLE. Frontal measurements included spectra from all available frontal lobe tissue while temporal lobe measurements were limited to a single voxel in the hippocampus. Prior studies have shown widespread neuron loss associated with partial epilepsy.‘,* Measurement of a larger volume of tissue may produce more robust results. From our findings we can not infer a cause-effect relationship between seizure frequency and NAA levels. Nevertheless, this study provides the first step toward understanding this relationship. Seizures may cause neuron loss or dysfunction resulting in decreased
Magnetic ResonanceImaging0 Volume 15, Number 4, 1997
418
NAA. Alternatively, patients with more extensive underlying neuron loss or dysfunction may be susceptible to having more seizures. Serial NAA studies controlled for seizure frequency should enable us to address the issue of causality. are gratefulto JohnWalker,PhD and WilliamJ. Marks,Jr., M.D. for theirhelpfulsuggestions. Dr. Laxeris supported by NIH grantROl-NS31966.
Acknowledgements-We
REFERENCES 1. Callaghan,N.; Garrett, A.; Goggin, T. Withdrawal of anticonvulsantdrugsin patientsfree of seizuresfor two years.N. Engl. J. Med. 318:942-946; 1988. 2. Dodson,W.E. Quality of Life Measurementsin Children with Epilepsy. In: M.R. Trimble and W.E. Dodson, (Eds) . Epilepsy and Quality of Life. New York: Raven Press,Ltd.; 1994: 219. 3. Hugg, J.W.; Laxer, K.D.; Matson, G.B.; Maudsley, A.A.; Weiner, M.W. Neuron loss localizes human focal epilepsy by in vivo proton MR spectroscopicimaging.Ann. Neurol. 34:788-794; 1993.
4. Hugg, J.W.; Duijn, J.H.; Matson, G.B.; Maudsley, A.A.; Tsuruda,J.S.; Gelinas,D.F.; Weiner,M.W. Elevatedlactate and alkalosisin chronic humanbrain infarction observed by 1H and 31P MR spectroscopicimaging. J. Cereb.Blood Flow Metab. 12:734-744; 1992. 5. Arnold, D.L.; Shoubridge,E.A.; Villemure, J.G.; Feindel, W. Proton andphosphorusmagneticresonancespectroscopy of humanastrocytomasin vivo. Preliminary observations on tumor grading. NMR Biomed. 3:184- 189; 1990. 6. Ebisu, T.; Rooney, W.D.; Graham,S.H.; Weiner, M.W.; Maudsley, A.A. N-acetyl-aspartateas an in vivo marker of neuronalviability in kainate-inducedstatusepilepticus: ‘H magneticresonancespectroscopicimaging. J. Cereb. Blood Flow Metab. 14:373-382; 1994. 7. Garcia,P.A.; Laxer, K.D.; van der Grond, J.; Hugg,J.W.; Mattson, G.B.; Weiner, M.W. Proton magneticresonance spectroscopicimaging in patientswith frontal lobe epilepsy. Ann. Neurol. 37:279-281; 1995. 8. Cendes,F.; Andermann,F.; Preul, MC.; Arnold, D.L. Lateralizationof temporallobe epilepsybasedon regional metabolic abnormalitiesin proton magnetic resonance spectroscopicimages.Ann. Neurol. 35:211-216; 1994.