SEARCH

SEARCH BY CITATION

Keywords:

  • Epilepsy surgery;
  • Seizure localization;
  • Epilepsy imaging

Summary

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References

Purpose:  Interictal positron emission tomography (PET) and ictal subtraction single photon emission computed tomography (SPECT) of the brain have been shown to be valuable tests in the presurgical evaluation of epilepsy. To determine the relative utility of these methods in the localization of seizure foci, we compared interictal PET and ictal subtraction SPECT to subdural and depth electrode recordings in patients with medically intractable epilepsy.

Methods:  Between 2003 and 2009, clinical information on all patients at our institution undergoing intracranial electroencephalography (EEG) monitoring was charted in a prospectively recorded database. Patients who underwent preoperative interictal PET and ictal subtraction SPECT were selected from this database. Patient characteristics and the findings on preoperative interictal PET and ictal subtraction SPECT were analyzed. Sensitivity of detection of seizure foci for each modality, as compared to intracranial EEG monitoring, was calculated.

Key Findings:  Fifty-three patients underwent intracranial EEG monitoring with preoperative interictal PET and ictal subtraction SPECT scans. The average patient age was 32.7 years (median 32 years, range 1–60 years). Twenty-seven patients had findings of reduced metabolism on interictal PET scan, whereas all 53 patients studied demonstrated a region of relative hyperperfusion on ictal subtraction SPECT suggestive of an epileptogenic zone. Intracranial EEG monitoring identified a single seizure focus in 45 patients, with 39 eventually undergoing resective surgery. Of the 45 patients in whom a seizure focus was localized, PET scan identified the same region in 25 cases (56% sensitivity) and SPECT in 39 cases (87% sensitivity). Intracranial EEG was concordant with at least one study in 41 cases (91%) and both studies in 23 cases (51%). In 16 (80%) of 20 cases where PET did not correlate with intracranial EEG, the SPECT study was concordant. Conversely, PET and intracranial EEG were concordant in two (33%) of the six cases where the SPECT did not demonstrate the seizure focus outlined by intracranial EEG. Thirty-three patients had surgical resection and >2 years of follow-up, and 21 of these (64%) had Engel class 1 outcome. No significant effect of imaging concordance on seizure outcome was seen.

Significance:  Interictal PET and ictal subtraction SPECT studies can provide important information in the preoperative evaluation of medically intractable epilepsy. Of the two studies, ictal subtraction SPECT appears to be the more sensitive. When both studies are used together, however, they can provide complementary information.

The cumulative lifetime incidence of epilepsy has been reported to be as high as 2–4% of the population (Hopkins & Shorvon, 1995). Although the majority of these patients will respond to antiepileptic medications, about 20–40% of them will continue to experience debilitating seizures despite maximal medical therapy (Murphy et al., 2001). Epilepsy surgery offers the potential for seizure control in a subgroup of these patients (Engel, 1996; Murphy et al., 2001), although this usually necessitates an accurate determination of the epileptic focus (Spencer, 1994; Engel, 1996). The current gold standard for seizure localization is intracranial epilepsy monitoring with subdural and/or intracerebral depth electrodes (Spencer, 1994). The inherent invasiveness of this method, however, has fueled the development of several noninvasive techniques that can assist with precise anatomic localization of seizure onset, potentially allowing for less-extensive intracranial monitoring arrays, or obviating the need for intracranial recording altogether.

The application of magnetic resonance imaging (MRI) in epilepsy has revolutionized the treatment of mesial temporal sclerosis (MTS) and lesional epilepsy (Won et al., 1999; Zaknun et al., 2008). By the same token, the development of functional neuroimaging with interictal 18F-fluorodeoxygluocose–positron emission tomography (FDG-PET), interictal single photon emission computed tomography (SPECT), ictal SPECT, or ictal subtraction SPECT (interictal SPECT fused, normalized and subtracted from ictal SPECT) has had a significant impact on the investigation of nonlesional epilepsy (Jack et al., 1994; Spencer, 1994; Cendes et al., 1997; Connelly et al., 1998; O’Brien et al., 1998, 1999, 2000; Stanley et al., 1998; Hwang et al., 2001; Lee et al., 2001a,b; Murphy et al., 2001; Henry & Van Heertum, 2003; la Fougère et al., 2009). State-of-the-art imaging with these techniques has arguably become an integral part of a multimodality imaging platform, and it has been shown that their use can lead to fewer invasive electroencephalography (EEG) recordings (Jack et al., 1994; Hwang et al., 2001). Several studies have described the utility, including the sensitivity and specificity, of these techniques when used individually (Jack et al., 1994; Spencer, 1994; O’Brien et al., 1998, 1999, 2000; Stanley et al., 1998; Hwang et al., 2001; Lee et al., 2001a,b; Murphy et al., 2001; Henry & Van Heertum, 2003; la Fougère et al., 2009), but there remain few data on the utility of their combined use in the same patients (Stefan et al., 1987; Coubes et al., 1993; Won et al., 1999; Hwang et al., 2001). Most of these studies have focused mostly on temporal lobe epilepsy and have compared positron emission tomography (PET) to either ictal SPECT or interictal SPECT, and not to ictal subtraction SPECT (Lewis et al., 2000). In addition, they have been conducted at a relatively early stage in the development of these imaging techniques. Comparative analysis of these imaging methods has remained difficult because of their limited availability, their cost, and rapid technologic advances.

In the current study, we compared the sensitivity of ictal subtraction SPECT and interictal PET, and the utility of their combined use, in patients with medically refractory epilepsy, by comparing them to the localization of epileptogenic cortex via intracranial EEG.

Methods

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References

Patient population

We retrospectively reviewed a prospectively collected database of patients that underwent intracranial epilepsy monitoring at Dartmouth-Hitchcock Medical Center (Lebanon, New Hampshire) between 2003 and 2009. Patients who had undergone both preoperative interictal PET and ictal subtraction SPECT as well as invasive monitoring with depth and/or subdural electrodes, were selected from this database. All patients had undergone a prior MRI study demonstrating no tumors or vascular lesions that were thought to account for the seizures. Patient baseline characteristics, findings on preoperative interictal PET and ictal subtraction SPECT, findings on intracranial EEG, and subsequent management were analyzed. For patients undergoing surgical resection, seizure outcomes of those with 24 months or greater follow-up were reviewed, and the effect of agreement of SPECT, PET, and intracranial EEG was analyzed.

Interictal PET

PET scans were performed on a General Electric Discovery ST machine (General Electric Company, Fairfield, CT, U.S.A.) 60 min after the intravenous administration of FDG in a dose of 5.55 mBq/kg (0.15 mCi/kg). Acquisitions were obtained over 8 min in the three-dimensional (3D) mode. Attenuation correction was provided by a computed tomography (CT) scan acquired at 120 kV and 140 mA for 4.9 s. If motion occurred and was deemed deleterious to the scan by the reviewing physician, the imaging was repeated. The data were reconstructed with ordered subset expectation maximization (OSEM). The PET was usually performed on an outpatient basis, with the patients showing no clinical signs of seizures. EEG was not performed during uptake of FDG. The PET data were interpreted by a nuclear medicine specialist (A.S.).

Ictal subtraction SPECT

Patients were admitted to the inpatient video-EEG monitoring unit at Dartmouth-Hitchcock Medical Center (Lebanon, New Hampshire) and if necessary had their antiepileptic medications withdrawn sequentially to precipitate seizures. The seizures were detected by direct observation and computerized scalp EEG detection devices. A nurse or technician supervised the video-EEG recording and sat continuously with the patient from 9 a.m. to 3 p.m. on days when the ictal SPECT was to be attempted. A shielded syringe containing technetium-99m ethyl cysteinate dimer (Neurolite; Bristol Myers Squibb, New Brunswick, NJ, U.S.A.) was brought to the unit and kept by the patient’s bedside. This syringe was recalibrated in the nuclear medicine department every 2 h, resulting in an injected dose of 925–1,110 MBq (25–30 mCi). Injection of radiotracer was commenced as rapidly as possible, within 5–30 s after seizure onset, which was detected by either clinical signs or EEG changes, whichever came first. Two or more seizures (a typical seizure type for that particular patient), in addition to a baseline interictal state, were recorded by continuous EEG for each patient. Most patients had only one ictal and one interictal SPECT scan, but two patients had second ictal scans. In these two instances, both ictal scans were reviewed by the multidisciplinary team and the scan with the greatest clinical correlation was selected.

Patients were transferred to nuclear medicine for imaging following clinical stabilization. All scans were acquired on a triple-headed gamma camera using high-resolution low-energy collimators. Four consecutive 5-min acquisitions were performed, each with 40 stops per head at 3 degrees per stop in a 128 × 128 matrix. The raw data files were summed before image processing after confirming the lack of patient movement. If movement occurred files were omitted as appropriate (maximum of 2 omitted). All studies were reconstructed in a similar manner using a ramp filter followed by postprocessing with a low pass filter and then underwent attenuation correction with the Chang technique. Studies were reconstructed in the axial, sagittal, and coronal planes, with axial slices one pixel thick (2.225 mm) used for registration purposes. The MRI studies were acquired on a General Electric Sigma 1.5 T magnet. As part of the routine epilepsy protocol, a 3D spoiled gradient recalled (SPGR) sequence was obtained in the coronal plane and utilized for registration purposes.

Image registration was performed with a fully automated algorithm (RView) with a technique described previously, creating a registered set including the MRI, interictal and ictal scans, and a fusion of the ictal–interictal subtraction and the MRI (Avery et al., 2000a,b; Lewis et al., 2000). Image subtraction was performed using Rview® utilizing a fully automated technique. No additional image smoothing was performed beyond the filters used in the reconstruction of the original SPECT images. Ictal SPECT was normalized to the interictal SPECT based upon global count rate. Subtraction images were viewed routinely by reduce the upper level of the image window to 25% of the maximum of the normalized SPECT images. Images were interpreted by the senior author (A.S), a nuclear medicine specialist. Seizure foci were identified if visible on the comparison of the ictal and interictal perfusion scans and confirmed on the subtraction image with clinical history taken into account. Confirmation of a discrete region of hyperperfusion was made if the area identified on the subtraction image was discernible from background and the most intense area of activity within the processed scan. Z-score maps were not used.

Intracranial EEG monitoring and surgical resection

Intracranial EEG was done with varying combinations of 12-contact depth electrodes, platinum 1 × 8 or 2 × 8 strip electrodes, and 4 × 8 or 8 × 8 grid electrodes. A specially designed 3 × 8 two-sided grid electrode was used for interhemispheric recordings (AdTech Medical Instrument Corporation, Racine, WI, U.S.A.). Up to 128 channels of intracranial EEG were recorded referentially and analyzed in referential and bipolar montages using Telefactor® equipment (Grass Technologies, West Warwick, RI, U.S.A.). Patients had their antiepileptic medications tapered, and were monitored until they experienced at least three of their stereotypical seizures. Decisions about the nature of the intracranial monitoring arrays were made by a multidisciplinary team based upon the preoperative data, including SPECT and PET findings. Subsequent decisions about the intended surgical resection were later made by the same team.

Results

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References

Demographics

Fifty-three patients underwent intracranial EEG monitoring with preoperative interictal PET and ictal subtraction SPECT scans. The average patient age was 32.7 years (mean 32 years, range 1–60 years), and four patients were younger than 18 years of age. Twenty-seven patients were female (Table 1).

Table 1.   Imaging findings and outcomes
Patient no.AgeSex (M/F)MRI findingsSubtraction SPECT abnormalityPET abnormalityLocalization on intracranial EEGSurgeryPathologyFollow-up (months)Engel outcome
 134MRight frontal venous malformationRight temporal and insulaRight anterior temporalRight mesial temporalRight selective amygdalohippocampectomyNormal36I
 249FNormalLeft temporal neocortexNoneNoneVNSN/A  
 320MNormalLeft posterior frontalNoneNoneNoneN/A  
 440MNormalLeft frontalNoneLeft frontalLeft frontal resectionCortical dysplasia36I
 537FRight MTSRight temporalRight temporalRight anterior temporalRight anterior temporal resectionMesial temporal sclerosis72I
 636MRight MTSLeft temporal and insulaLeft temporalLeft mesial temporalLeft selective amygdalohippocampectomyMesial temporal sclerosis0 
 741FNormalLeft temporalLeft temporalLeft mesial temporalLeft selective amygdalohippocampectomyMesial temporal sclerosis84I
 829FNormalRight frontalNoneLeft medial frontalLeft medial frontal resectionNormal39III
 933MNormalRight frontalRight parietalRight frontalRight frontal resectionNormal24IV
1034FNormalRight temporalRight temporalRight anterior temporalRight anterior temporal resectionNormal28I
1138FLeft MTSLeft temporalLeft temporalLeft temporalLeft selective amygdalohippocampectomyMesial temporal sclerosis63III
1227MNormalRight temporalNormalNoneVNSN/A  
1329FLeft MTSLeft temporalLeft temporalLeft mesial temporalLeft selective amygdalohippocampectomyMesial temporal sclerosis52I
1439FNormalLeft temporalLeft temporalLeft anterior temporalLeft anterior temporal resection sparing hippocampusCortical dysplasia99I
1529MPrevious corpus callosotomyLeft posterior frontalNormalLeft frontalLeft frontal resectionGliosis41III
1616MRight frontal cortical dysplasiaLeft frontalDiffuse right hemisphereRight frontalRight frontal resectionCortical dysplasia36I
1745MNormalLeft temporal/insula/frontalNormalLeft anterior frontalLeft frontal resectionNormal39I
1849FLeft insula and Right frontal increased T2 signalLeft frontalNormalLeft medial frontalLeft medial frontal resectionCortical dysplasia33II
1937FNormalLeft posterior frontal, Left insulaNormalLeft frontalLeft frontal resectionCortical dysplasia3 
2026FNormalRight temporalRight temporalRight temporalRight temporal resectionGliosis31I
2153FLeft frontal metal fragmentRight frontalRight frontalRight frontal and parietalNone   
2252MLeft high parietal gliosisLeft parietalNormalLeft parietalLeft parietal resectionGliosis94I
2325FLeft frontal FLAIR signalLeft frontoparietalNormalNoneNeuroPaceN/A  
2418MLeft hemisphere encephalomalaciaLeft parietalLeft parietalLeft frontoparietalLeft frontoparietal resectionEncephalomalacia22 
2526MLeft MTS and left frontal cavernomaLeft temporalLeft temporalLeft temporalLeft temporal lobectomyMesial temporal sclerosis57I
2645MNormalLeft temporalNormalBilateral mesial temporalNoneN/A  
273FMultiple bilateral cortical tubersRight temporal and parietalDiffuse bilateralRight frontoparietalRight frontoparietal resectionNormal1IV
2820FFocal dysplasia left hemisphereLeft parietalLeft parietal and frontalLeft inferior frontalMultiple subpial transectionsN/A  
2922MLeft cortical dysplasia with polymicrogyriaLeft posterior temporal, Right temporalLeft temporalLeft temporalAnterior corpus callosotomy and VNSN/A  
301MNormalLeft parietal, occipitotemporalDiffuse left hemisphereLeft occipitotemporalLeft occipitotemporal resectionCortical dysplasia33III
3149FRight hippocampal sclerosisRight medial temporalNormalRight mesial temporalRight selective hippocampectomyMesial temporal sclerosis38I
3242FRight parietal increased FLAIR intensityRight medial frontalNormalRight medial frontalRight medial frontal resectionCortical dysplasia72I
3321MNormalLeft frontalNormalLeft medial frontalLeft medial frontal resectionNormal39III
3416MNormalLeft frontalNormalLeft frontalLeft frontal resectionNot sent5 
3529MNormalLeft frontalNormalNoneCorpus callosotomyN/A  
3643MRight MTSRight temporalRight temporalRight mesial temporalRight selective hippocampectomyMesial temporal sclerosis65I
3723FNormalRight posterior temporal lobe and insulaRight parietalRight medial frontalRight SMA resectionNormal67II
3832MRight frontal gyrus dysplasiaRight temporalNormalRight parietalVNSN/A  
3957FLeft hippocampal atrophyLeft temporalDiffuse left hemisphereLeft mesial temporalLeft standard temporal lobectomyMesial temporal sclerosis69I
4039MNormalRight temporal and parietalNormalBilateral mesial temporalNoneN/A  
4137FLeft hippocampal atrophyRight temporalLeft temporalLeft mesial temporalLeft selective hippocampectomyMesial temporal sclerosis79II
4226MSurgical changes from previous callosotomyBilateral frontalNormalNoneVNSN/A  
4349FT2 signal abnormality in left frontal gyrusLeft frontalNormalLeft frontalMultiple subpial transectionN/A  
4430MLeft frontal periventricular white matter changesRight frontalNormalRight frontalRight frontal resectionNot sent24I
4522FNormalRight frontalNormalLeft orbitofrontalLeft orbitofrontal resectionNot sent39II
4660MIncreased T2 prolongation in left hippocampusRight temporalNormalRight temporalRight temporal lobectomyNot sent88I
4725MLeft frontal encephalomalaciaLeft frontotemporalLeft frontalRight temporalNoneN/A  
4847MNormalRight insula and temporalRight temporalRight temporalRight temporal lobectomyEncephalomalacia30II
4921FPrevious frontal craniotomyLeft frontalLeft frontalLeft anterior frontalLeft frontal resectionCortical dysplasia31I
5029FRight temporal gliosisRight temporal and right frontalRight temporalRight temporalRight temporal lobectomyGliosis10 
5145FRight hippocampal sclerosisRight temporalRight temporalRight temporalRight temporal lobectomyMesial temporal sclerosis35II
5212FNormalLeft temporalLeft temporalLeft temporalRight temporal lobectomyGangliocytoma46I
5326FNormalRight temporalNormalRight temporalRight temporal lobectomyGliosis45I

Imaging

Twenty-three patients had no abnormal MRI findings. Ten patients had hippocampal atrophy or sclerosis. Seven patients had nonspecific regions of T2 prolongation. Six patients had regions suggestive of cortical dysplasia. Five patients had postsurgical changes or encephalomalacia; one had tuberous sclerosis and one had a developmental venous abnormality. Twenty-four patients had regions of relative hypometabolism on interictal PET. Twenty-two of these patients demonstrated a specific unilateral focus. A temporal focus was seen in 16 patients, a frontal focus in 3 patients, and a parietal focus in a further 3 patients. All 53 patients who were studied demonstrated a region of relative hyperperfusion on ictal subtraction SPECT with characteristics typical for a potential ictal onset zone. The mean injection time from seizure onset was 17.8 s. Injections took 1–2 s to complete. A unilateral focus was seen in 52 patients. Imaging revealed a discrete region of hyperperfusion in the temporal lobe in 21 patients and frontal lobe in 16 patients. SPECT suggested an ictal onset in the parietal lobe in three patients, frontotemporal region in three patients, and temporoinsular region in three patients. Other regions of possible seizure onset identified by SPECT were temporoparieto-occipital (2), frontoparietal (1), frontoinsular (1), temporoparietal (1), and frontotemporoinsular (1) (Table 1).

Intracranial EEG monitoring

Intracranial EEG monitoring identified a single seizure focus in 45 patients, but failed to localize a seizure focus in the other eight patients. Seizures were localized to the temporal lobe in 24 patients, the frontal lobe in 17 patients, the parietal lobe in 1 patient, occipitotemporal region in 1 patient, and to the frontoparietal region in 2 patients. Thirty-nine patients, all with a seizure focus identified on intracranial monitoring, went on to undergo resective surgery.

Surgery

Of the 39 patients who underwent resection of a seizure focus, 20 were temporal lobe resections (mesial, neocortical, or both), 15 were frontal resections, and 2 were frontoparietal resections. One patient underwent a parietal resection and another patient a temporooccipital resection. Fourteen patients did not undergo a subsequent resection surgery, either due to failure to localize seizures, bilateral seizure localization, or seizure localization in the eloquent cortex. Four of these patients underwent vagus nerve stimulator placement, two underwent corpus callosotomy, two underwent multiple subpial transection, and one responsive neurostimulation. Five patients did not undergo any further surgery (Table 1).

Pathology

Of the 39 patients undergoing resection surgery, pathologic examination of resected tissue was carried out in 35 cases. Mesial temporal sclerosis was identified histologically in 11 cases, cortical dysplasia in 8 cases, and gliotic changes in 5 cases (Table 1). A gangliocytoma, not evident on MRI, was demonstrated histologically in one case undergoing temporal lobectomy. Eight patients had normal pathologic findings, and two patients had findings of encephalomalacia only. Fourteen patients had abnormal pathology discovered other than mesial temporal sclerosis or encephalomalacia. Of these patients, nine had a normal preoperative MRI.

Correlation of SPECT, PET, and intracranial EEG findings

Of the 45 patients in whom a seizure focus was localized with intracranial monitoring, PET scan identified the same region in 25 cases (56% sensitivity) and SPECT in 39 cases (87% sensitivity). Intracranial EEG was concordant with at least one study in 41 cases (91%) and both studies in 23 cases (51%). In 20 cases, the intracranial EEG did not correlate with the PET study. These instances (17 of 20) were largely due to nonlocalization by PET (i.e., a normal PET scan). In 16 (80%) of these 20 cases where PET did not correlate with intracranial EEG, the SPECT study was concordant. Conversely, in six cases the ictal subtraction SPECT did not demonstrate the seizure focus outlined by intracranial EEG. In these cases the failure of the SPECT was one of false localization. PET and intracranial EEG were concordant in two (33%) of the six cases where the SPECT did not demonstrate the seizure focus outlined by intracranial EEG.

In MRI-negative patients, SPECT was concordant with intracranial EEG in 16 (64%) of 25 cases, and PET was concordant with intracranial EEG in 9 (36%) of 25 cases. Both were concordant with EEG in 9 (36%) of 25 cases. In MRI-positive patients, SPECT was concordant with intracranial EEG in 35 (87.5%) of 45 cases and PET was concordant with intracranial EEG in 25 (62.5%) of 40 cases. Both were concordant with EEG in 22 (55%) of 40 cases.

In patients with temporal epilepsy (defined by intracranial EEG) SPECT was concordant with intracranial EEG in 21 (95%) of 22 cases and PET was concordant with intracranial EEG in 18 (82%) of 22 cases. Both were concordant with EEG in 17 (77%) of 22 cases. In patients with extratemporal epilepsy (defined by intracranial EEG) SPECT was concordant with intracranial EEG in 18 (78%) of 23 cases, and PET was concordant with intracranial EEG in 7 (30%) of 23 cases. Both were concordant with EEG in 6 (26%) of 23 cases.

Seizure outcomes after surgery

Thirty-three patients undergoing surgical resection of a seizure focus had >24 months of follow-up. Mean and median follow-up were 50.4 and 39 months, respectively (range 24–99 months). Nineteen of these patients were female, and mean age was 34.3 years (range 1–60, median 34). Eighteen patients had temporal resections, 13 had frontal resections, and the remaining two patients had parietal and temporooccipital resections (Table 1). Intracranial EEG agreed with both SPECT and PET in 16 cases, SPECT exclusively in 12 cases, and PET exclusively in two cases. Three patients underwent surgery after intracranial EEG did not agree with prior SPECT and PET imaging.

Engel class 1 outcome was achieved in 21 patients (64%), class 2 in 6 patients (18%), class 3 in 5 patients (15%), and class 4 in 1 patient (3%). No significant difference was seen in the number of patients achieving seizure freedom in the group with concordance of SPECT, intracranial EEG, and PET, compared to those with concordance of SPECT and intracranial EEG alone (p = 0.69; Fisher exact test) (Table 2). A representative case is shown in Figure 1.

Table 2.   Seizure outcomes in patients undergoing resection and with 2 years or greater follow-up
 IC EEG no concordance with PET or SPECTIC EEG- SPECT concordance onlyIC EEG-PET concordance onlyIC EEG concordance with both SPECT and PETAll patients
Engel Class 10    8 (66.7%)   1 (50%)     12 (75%)21 (64%)
Engel Class 22    1   1      26 (18%)
Engel Class 31    2   0      25 (15%)
Engel Class 40    1   0      01 (3%)
Total3   12   2     1633
image

Figure 1.   Coregistered images, columns from left to right: MRI, PET, interictal SPECT, ictal SPECT, ictal subtraction SPECT: Representative case: A 43-year-old man with medically intractable complex partial seizures since adolescence was referred to the Dartmouth Epilepsy Service. Ictal SPECT (Colors on the scales are graded relative to intensity counts per pixel. For the ictal subtraction SPECT values are normalized relative to the interictal scan.) revealed right temporal hyperperfusion relative to the interictal study. Other clusters seen were considered artifactual based on clinical data. Interictal PET revealed hypometabolism in the right temporal lobe. The patient underwent intracranial EEG monitoring with recording electrodes placed over both temporal and frontal lobes. Seizure onset was found to originate in the right temporal lobe. The patient went on to undergo amygdalohippocampectomy. He is currently 65 months following his surgery and has Engel class 1 seizure outcome.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References

Although intracranial EEG monitoring remains the gold standard (Spencer, 1994) for localization of seizure foci in patients with intractable epilepsy, newer technologies have been added to the armamentarium of the surgical epilepsy team and have arguably transformed the field. MRI has revolutionized the detection of surgically treatable lesions (Cascino et al., 1993; Sisodiya et al., 1996; Ho et al., 1998), whereas functional imaging, such as SPECT and PET, play a complementary role in the localization or lateralization of the epileptogenic focus, minimizing the need for invasive EEG monitoring or enabling smaller intracranial studies (Spencer, 1994; Kilpatrick et al., 1997). Several studies have analyzed the contribution of the various functional imaging techniques independently in seizure localization (Spencer, 1994; Knowlton et al., 1997; Theodore et al., 1997; Thadani et al., 2004). Few, however, have directly compared the various modalities (Stefan et al., 1987; Cendes et al., 1997; Won et al., 1999; Hwang et al., 2001).

According to a meta-analysis by Spencer (1994)), the highest sensitivity was seen with ictal SPECT (90% in temporal lobe epilepsy, 81% in extratemporal epilepsy). Among the interictal techniques, PET showed the highest sensitivity in patients with temporal lobe epilepsy (84% vs. 66% for interictal SPECT, 55% for qualitative MRI), whereas interictal SPECT showed the highest sensitivity in patients with extratemporal epilepsy (60% vs. 43% for MRI and 33% for PET). Markand et al. (1997) have shown that in their study population the sensitivity of ictal SPECT was almost equal to that of PET (86%) in patients with complex partial seizures. On the other hand, Won et al. (1999) reported that PET had the highest sensitivity and specificity for both temporal and extratemporal lesions. However, inconsistencies in the injection of the radiotracer during the ictal phase in this study might have contributed to the low sensitivity of SPECT scans. Ictal subtraction SPECT has been shown to increase the sensitivity and specificity of unsubtracted ictal SPECT (O’Brien et al., 1998, 1999, 2000). In the current analysis, the sensitivity of ictal subtraction SPECT is 87%, higher than in previously reported literature. This may be due to all injections for ictal SPECT in our series taking place <30 s from seizure onset (with a mean injection time of 17.8 s).

There are few studies comparing the results of the various functional imaging techniques (Stefan et al., 1987; Cendes et al., 1997; Won et al., 1999; Hwang et al., 2001). The nonconcordance rate among the different imaging techniques has been reported (Won et al., 1999) to be as high as 30–40%. The various modalities measure different aspects of the epileptic process, namely structure (MRI), metabolism (FDG-PET), and perfusion (SPECT). MRI is exceptional in delineating anatomic abnormalities that can contribute to epilepsy, whereas PET can provide imaging of the cerebral metabolism, although it is limited to the interictal phase. It takes approximately 1 h for the radiotracer to enter the cells to be metabolized and to be distributed throughout the brain tissue, making the imaging of the short ictal state impractical (Won et al., 1999). Indeed, if radiotracer is injected during a seizure and PET performed 1 h after the injection, the scan would likely reflect the interictal or postictal state. In contrast, SPECT can demonstrate cerebral blood flow changes in the interictal as well as in the ictal period. This is possible because the radiotracer takes approximately 1 min to be distributed throughout the brain and remains in the brain tissue after cessation of seizure activity without significant redistribution, allowing interictal scanning of the ictal state (Won et al., 1999). Although interictal SPECT has been used in the localization of seizure onset by demonstrating decreased regional cerebral perfusion, there is extensive evidence demonstrating it has a relatively low diagnostic yield. Subtracting interictal images from ictal scans and coregistering the final image with MRI appears to result in superior temporal and spatial localization of the epileptogenic focus.

In the present analysis, we demonstrated that intracranial EEG was concordant with at least one study in 91% of the patients and both studies in 51% of the patients. In this study, PET and SPECT were shown to provide complementary information. In 16 of 20 cases (80%) where PET did not correlate with intracranial EEG, the SPECT study was concordant. Conversely, PET and intracranial EEG were concordant in two of the six cases (33%) where the SPECT did not demonstrate the seizure focus outlined by intracranial EEG. Although in this study, their agreement did not show statistically improved seizure outcomes after resection, their use in combination increases sensitivity and appears to identify epileptogenic regions that may otherwise not be studied with intracranial EEG.

In the current study we also found increased concordance of both PET and SPECT findings with those from intracranial EEG in MRI-positive patients and those patients with temporal lobe epilepsy. It is possible that improved performance in MRI-positive patients may be the result of structural epileptogenic lesions being more susceptible to ictal hyperperfusion or to interictal hypometabolism, with a well-delineated anatomic seizure-onset zone showing good contrast from surrounding nonepileptogenic areas. The improved performance in TLE is not intuitive but may result from overall improved sensitivity of both ictal subtraction SPECT and interictal PET in TLE (Spencer, 1994).

It is well-recognized that patients with positive MRI findings and/or temporal lobe onset have relatively better seizure outcomes after surgery than other patients. Given the finding in our series of improved concordance of SPECT and PET with intracranial EEG in these patients, one might expect overall improved seizure outcomes in the patients with SPECT and PET concordance with intracranial EEG. We found, however, no statistical difference in 2-year seizure outcome after resection in patients with concordance of SPECT, PET, and EEG compared to SPECT and EEG alone. It is possible that a larger sample of patients may have detected such a difference.

In our study, when compared to intracranial EEG recording, neither SPECT nor PET achieved 100% sensitivity or specificity. The effect of seizure pathophysiology on cerebral perfusion and metabolism remains incompletely understood, Potential reasons for failure to detect seizures or false localization include rapid seizure propagation, late injection of radiotracer, or if the seizures are not of sufficient magnitude to produce distinguishable changes in blood flow or metabolism relative to adjacent tissue using the resolution of current scanners.

The present study has limitations that should be considered. It is retrospective in nature and the patients analyzed are limited to a single center. Although all patients had medically intractable epilepsy without a clear structural cause after video-scalp EEG monitoring, and MRI, SPECT, and PET, imaging, they nonetheless comprised a heterogenous group with respect to age, seizure history, MRI findings, specific types of intracranial electrode implantation, and eventual pathology. In addition, the patients in this study did not undergo magnetoencephalography (MEG), a commonly used preoperative seizure localization adjunct in many centers, and therefore we cannot speculate on its additional utility. Similarly, our PET system did not include partial volume correction, which can be of value in detection of seizure foci in patients with epilepsy Finally, because there is no control group in this study, it cannot answer whether the addition of PET or SPECT improves eventual seizure outcomes after surgery.

Conclusions

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References

Interictal PET and ictal subtraction SPECT studies can provide important information in the preoperative evaluation of medically intractable epilepsy. Of the two, ictal subtraction SPECT appears to be the most sensitive in the identification of an epileptogenic focus. Their combined use, however, appears to lead to the identification of more epileptogenic regions, that are subsequently confirmed by intracranial EEG. Therefore, both imaging modalities can be important, complementary adjuncts in the investigation of patients with inconclusive localization on video-EEG and MRI.

Disclosure

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References

None of the authors has any conflict of interest to disclose. We confirm that we have read the journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

References

  1. Top of page
  2. Summary
  3. Methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Disclosure
  8. References
  • Avery RA, Spencer SS, Studholme C, Stokking R, Morano G, Corsi M, Seibyl JP, Spencer DD, Zubal IG. (2000a) Reproducibility of serial peri-ictal single-photon emission tomography difference images in epilepsy patients undergoing surgical resection. Eur J Nucl Med27:5055.
  • Avery RA, Zubal IG, Stokking R, Studholme C, Corsi M, Seibyl JP, Spencer SS. (2000b) Decreased cerebral blood flow during seizures with ictal SPECT injections. Epilepsy Res40:5361.
  • Cascino GD, Jack CRJ, Sharbrough FW, Kelly PJ, Marsh WR. (1993) MRI assessments of hippocampal pathology in extratemporal lesional epilepsy. Neurology43:23802382.
  • Cendes F, Caramanos Z, Andermann F, Dubeau F, Arnold DL. (1997) Proton magnetic resonance spectroscopic imaging and magnetic resonance imaging volumetry in the lateralization of temporal lobe epilepsy: a series of 100 patients. Ann Neurol42:737746.
  • Connelly A, Van Paesschen W, Porter DA, Johnson CL, Duncan JS, Gadian DG. (1998) Proton magnetic resonance spectroscopy in MRI-negative temporal lobe epilepsy. Neurology51:6166.
  • Coubes P, Awad IA, Antar M, Magdinec M, Sufka B. (1993) Comparison and spacial correlation of interictal HMPAO-SPECT and FDG-PET in intractable temporal lobe epilepsy. Neurol Res15:160168.
  • Engel JJ. (1996) Surgery for seizures. N Engl J Med334:647652.
  • Henry TR, Van Heertum RL. (2003) Positron emission tomography and single photon emission computed tomography in epilepsy care. Semin Nucl Med33:88104.
  • Ho SS, Kuzniecky RI, Gilliam F, Faught E, Morawetz R. (1998) Temporal lobe developmental malformations and epilepsy: dual pathology and bilateral hippocampal abnormalities. Neurology50:748754.
  • Hopkins A, Shorvon S. (1995) Definitions and epidemiology of epilepsy. In Hopkins A, Shorvon S, Cascino G (Eds) Epilepsy. Chapman and Hall, London, pp. 124.
  • Hwang SI, Kim JH, Park SW, Han MH, Yu IK, Lee SH, Lee DS, Lee SK, Chung CK, Chang KH. (2001) Comparative analysis of MR imaging, positron emission tomography, and ictal single-photon emission CT in patients with neocortical epilepsy. AJNR Am J Neuroradiol22:937946.
  • Jack CRJ, Mullan BP, Sharbrough FW, Cascino GD, Hauser MF, Krecke KN, Luetmer PH, Trenerry MR, O’Brien PC, Parisi JE. (1994) Intractable nonlesional epilepsy of temporal lobe origin: lateralization by interictal SPECT versus MRI. Neurology44:829836.
  • Kilpatrick C, Cook M, Kaye A, Murphy M, Matkovic Z. (1997) Non-invasive investigations successfully select patients for temporal lobe surgery. J Neurol Neurosurg Psychiatry63:327333.
  • Knowlton RC, Laxer KD, Ende G, Hawkins RA, Wong ST, Matson GB, Rowley HA, Fein G, Weiner MW. (1997) Presurgical multimodality neuroimaging in electroencephalographic lateralized temporal lobe epilepsy. Ann Neurol42:829837.
  • la Fougère C, Rominger A, Förster S, Geisler J, Bartenstein P. (2009) PET and SPECT in epilepsy: a critical review. Epilepsy Behav15:5055.
  • Lee DS, Lee JS, Kang KW, Jang MJ, Lee SK, Chung JK, Lee MC. (2001a) Disparity of perfusion and glucose metabolism of epileptogenic zones in temporal lobe epilepsy demonstrated by SPM/SPAM analysis on 15O water PET, [18F]FDG-PET, and [99mTc]-HMPAO SPECT. Epilepsia42:15151522.
  • Lee DS, Lee SK, Lee MC. (2001b) Functional neuroimaging in epilepsy: FDG PET and ictal SPECT. J Korean Med Sci16:689696.
  • Lewis PJ, Siegel A, Siegel A, Studholme C, Sojkova J, Roberts DW, Thadani VM, Gilbert KL, Darcey TM, Williamson PD. (2000) Does performing image registration and subtraction in ictal brain SPECT help localize neocortical seizures?J Nucl Med41:16191626.
  • Markand ON, Salanova V, Worth R, Park HM, Wellman HN. (1997) Comparative study of interictal PET and ictal SPECT in complex partial seizures. Acta Neurol Scand95:129136.
  • Murphy M, O’Brien TJ, Morris K, Cook MJ. (2001) Multimodality image-guided epilepsy surgery. J Clin Neurosci8:534538.
  • O’Brien TJ, So EL, Mullan BP, Hauser MF, Brinkmann BH, Bohnen NI, Hanson D, Cascino GD, Jack CRJ, Sharbrough FW. (1998) Subtraction ictal SPECT co-registered to MRI improves clinical usefulness of SPECT in localizing the surgical seizure focus. Neurology50:445454.
  • O’Brien TJ, So EL, Mullan BP, Hauser MF, Brinkmann BH, Jack CRJ, Cascino GD, Meyer FB, Sharbrough FW. (1999) Subtraction SPECT co-registered to MRI improves postictal SPECT localization of seizure foci. Neurology52:137146.
  • O’Brien TJ, So EL, Mullan BP, Cascino GD, Hauser MF, Brinkmann BH, Sharbrough FW, Meyer FB. (2000) Subtraction peri-ictal SPECT is predictive of extratemporal epilepsy surgery outcome. Neurology55:668677.
  • Sisodiya SM, Stevens JM, Fish DR, Free SL, Shorvon SD. (1996) The demonstration of gyral abnormalities in patients with cryptogenic partial epilepsy using three-dimensional MRI. Arch Neurol53:2834.
  • Spencer SS. (1994) The relative contributions of MRI, SPECT, and PET imaging in epilepsy. Epilepsia35:S72S89.
  • Stanley JA, Cendes F, Dubeau F, Andermann F, Arnold DL. (1998) Proton magnetic resonance spectroscopic imaging in patients with extratemporal epilepsy. Epilepsia39:267273.
  • Stefan H, Pawlik G, Böcher-Schwarz HG, Biersack HJ, Burr W, Penin H, Heiss WD. (1987) Functional and morphological abnormalities in temporal lobe epilepsy: a comparison of interictal and ictal EEG, CT, MRI, SPECT and PET. J Neurol234:377384.
  • Thadani VM, Siegel A, Lewis P, Siegel AM, Jobst BC, Gilbert KL, Darcey TM, Roberts DW, Williamson PD. (2004) Validation of ictal single photon emission computed tomography with depth encephalography and epilepsy surgery. Neurosurg Rev27:2733.
  • Theodore WH, Sato S, Kufta CV, Gaillard WD, Kelley K. (1997) FDG-positron emission tomography and invasive EEG: seizure focus detection and surgical outcome. Epilepsia38:8186.
  • Won HJ, Chang KH, Cheon JE, Kim HD, Lee DS, Han MH, Kim IO, Lee SK, Chung CK. (1999) Comparison of MR imaging with PET and ictal SPECT in 118 patients with intractable epilepsy. AJNR Am J Neuroradiol20:593599.
  • Zaknun JJ, Bal C, Maes A, Tepmongkol S, Vazquez S, Dupont P, Dondi M. (2008) Comparative analysis of MR imaging, ictal SPECT and EEG in temporal lobe epilepsy: a prospective IAEA multi-center study. Eur J Nucl Med Mol Imaging35:107115.