Localizing value of ictal SPECT is comparable to MRI and EEG in children with focal cortical dysplasia

Authors

  • Pavel Krsek,

    1. Department of Pediatric Neurology, Charles University, 2nd Faculty of Medicine, University Hospital Motol, Prague, Czech Republic
    Search for more papers by this author
  • Martin Kudr,

    1. Department of Pediatric Neurology, Charles University, 2nd Faculty of Medicine, University Hospital Motol, Prague, Czech Republic
    Search for more papers by this author
  • Alena Jahodova,

    1. Department of Pediatric Neurology, Charles University, 2nd Faculty of Medicine, University Hospital Motol, Prague, Czech Republic
    Search for more papers by this author
  • Vladimir Komarek,

    1. Department of Pediatric Neurology, Charles University, 2nd Faculty of Medicine, University Hospital Motol, Prague, Czech Republic
    Search for more papers by this author
  • Bruno Maton,

    1. Department of Neurology and Comprehensive Epilepsy Program, Brain Institute, Miami Children's Hospital, Miami, Florida, U.S.A
    Search for more papers by this author
  • Stephen Malone,

    1. Department of Neurology and Comprehensive Epilepsy Program, Brain Institute, Miami Children's Hospital, Miami, Florida, U.S.A
    Search for more papers by this author
  • Ian Miller,

    1. Department of Neurology and Comprehensive Epilepsy Program, Brain Institute, Miami Children's Hospital, Miami, Florida, U.S.A
    Search for more papers by this author
  • Prasanna Jayakar,

    1. Department of Neurology and Comprehensive Epilepsy Program, Brain Institute, Miami Children's Hospital, Miami, Florida, U.S.A
    Search for more papers by this author
  • Trevor Resnick,

    1. Department of Neurology and Comprehensive Epilepsy Program, Brain Institute, Miami Children's Hospital, Miami, Florida, U.S.A
    2. Department of Neurology, University of Miami Miller School of Medicine, Miami, Florida, U.S.A
    Search for more papers by this author
  • Michael Duchowny

    Corresponding author
    1. Department of Neurology and Comprehensive Epilepsy Program, Brain Institute, Miami Children's Hospital, Miami, Florida, U.S.A
    2. Department of Neurology, University of Miami Miller School of Medicine, Miami, Florida, U.S.A
    • Department of Pediatric Neurology, Charles University, 2nd Faculty of Medicine, University Hospital Motol, Prague, Czech Republic
    Search for more papers by this author

Address correspondence to Michael Duchowny, Department of Neurology, Brain Institute, Miami Children's Hospital, University of Miami Miller School of Medicine, 3200 S.W. 60th Court, Miami, FL, U.S.A. E-mail: michael.duchowny@mch.com

Summary

Purpose

To assess the predictive value of ictal single-photon emission computed tomography (SPECT) for outcome after excisional epilepsy surgery in a large population of children with focal cortical dysplasia (FCD).

Methods

One hundred seventy-three ictal SPECT studies in 106 children with histologically proven FCD were retrospectively analyzed. The extent and location of ictal hyperperfusion and completeness of surgical removal were assessed. Completeness of resection of epileptogenic regions defined by ictal SPECT, electroencephalography (EEG), and magnetic resonance imaging (MRI) were compared and correlated with postoperative seizure outcome. In addition, subcortical activation of the cerebellum, basal ganglia, and thalamus were analyzed.

Key Findings

The extent of hyperperfusion was focal or lobar in 58%, whereas multilobar activations occurred in only 32%; hemispheric or bilateral findings were rare. Favorable postsurgical seizure outcome was achieved in 67% patients with nonlocalized SPECT findings, 45% with nonresected ictal hyperperfusion, 36% with partially resected ictal hyperperfusion, and 86% when the zone of ictal hyperperfusion was completely resected (p = 0.000198). The favorable postsurgical outcome after complete removal of the SPECT hyperperfusion zone surpassed the 75% rate of seizure freedom in patients with removal of MRI/EEG-defined epileptogenic region. A similar predictive value of ictal SPECT for seizure outcome was found in nonoperated patients and subjects who were undergoing reoperation. Subcortical activation conferred no predictive value.

Significance

Ictal SPECT helps to define the epileptogenic zone in a high proportion of children with FCD undergoing surgical evaluation. Complete removal of both SPECT and MRI/EEG-defined regions is a strong predictor of surgical success and has important implications for surgical planning.

Ictal single-photon emission computed tomography (SPECT) is a noninvasive neuroimaging method for localizing seizure onset in focal epileptic seizures. Previous studies reveal that regions of cortical hyperperfusion assist in guiding the location of excisional epilepsy surgery (Buchhalter & So, 2004; Van Paesschen, 2004; Van Paesschen et al., 2007). When SPECT localization is concordant with the site of the surgical excision, surgical outcome is frequently favorable (O'Brien et al., 1998, 2000; So, 2000; Kaminska et al., 2003; Wetjen et al., 2006). However, the value of ictal SPECT in comparison to other localizing techniques remains unresolved. For example, removal of hyperperfused regions visualized by ictal/interictal subtraction images coregistered with magnetic resonance imaging (MRI) (SISCOM focus) is associated with higher rates of postoperative seizure outcome (O'Brien et al., 2000, 2004; Cascino et al., 2004), and it remains unclear whether the extent of cortical hyperperfusion alone is a reliable guide for planning the extent of surgical resection. The main concern is differentiating the primary seizure focus from regions of ictal propagation (Laich et al., 1997; Dupont et al., 2006; Huberfeld et al., 2006).

Several studies suggest that ictal SPECT findings predict postsurgical seizure outcome independent of MRI and scalp EEG data (Won et al., 1999; O'Brien et al., 2000; Zaknun et al., 2008). However, the comparative value of SPECT and MRI–electroencephalography (EEG) localization of the epileptogenic zone and the importance of complete resection of regions defined by individual diagnostic tests have not been examined in detail. Furthermore, subcortical activations in the cerebellum, basal ganglia, and thalami and their practical relevance to epilepsy surgery planning are unknown (Won et al., 1999; Shin et al., 2001; Sojkova et al., 2003; Van Paesschen et al., 2003).

Prior studies frequently combined pediatric and adult patients with diverse etiologies such as focal and hemispheric cortical malformations, tumors, tuberous sclerosis, or hippocampal sclerosis (Lawson et al., 2000; Kaminska et al., 2003). Focal cortical dysplasia (FCD) constitutes the most important cause of focal intractable epilepsy in childhood (Krsek et al., 2008). These patients represent a diagnostic and treatment challenge due to poorer postsurgical seizure outcomes compared to patients undergoing surgery for focal lesional epilepsy syndromes (Harvey et al., 2008). The value of colocalization of the cortical hyperperfusion zones with MRI-detected dysplastic lesions has been reported in adults (O'Brien et al., 2004; Dupont et al., 2006), but only one study focused exclusively on pediatric patients with FCD (Gupta et al., 2004). Pediatric patients with FCD are therefore an important but understudied group for defining the clinical usefulness of ictal SPECT.

To address these issues we analyzed a large cohort of children with histologically proven mild malformation of cortical development (mMCD) and FCD. The ultimate goal of our study was to enhance their noninvasive presurgical diagnosis and surgical management.

Methods

Inclusion criteria

Pediatric patients who underwent epilepsy surgery at the Miami Children's Hospital were retrospectively reviewed. We included only subjects who had (1) at least one excisional epilepsy surgery in Miami Children's Hospital, (2) definite histologic diagnosis of mMCD/FCD, (3) known seizure outcome at 2 years after (the last) surgery, and (4) at least one presurgical ictal SPECT study.

SPECT injection and imaging methods

Periictal injections of radioisotope were performed by specially trained registered radiology technicians during typical seizures while the patient was undergoing video-EEG monitoring. The radiotracer 99mTc-hexamethyl propylene amine oxime (HMPAO) was prepared at the Miami Children's Hospital Department of Radiology, Florida. Pediatric doses were calculated according to the patient's weight. Injections were administered as soon as either clinical (with EEG confirmation) or EEG seizure onset of a habitual seizure was observed. The radiotracer injection was followed by a saline flush. SPECT images were acquired within 2–3 h of the radiotracer injection. The acquisition was performed on a three-headed Multispect 3 Siemens' Medical System machine (Hoffman Estate, IL, U.S.A.) with the following acquisition parameters: 120 word mode, 360-degree rotations, 40 stops 120 images, 60 s per frame imaging, 1.28 magnification factor, fan beam collimation. A standard series of axial, coronal, and sagittal images was created. Selected patients had a pediatric sedation protocol during image acquisition. Orally administered chloral hydrate (50–75 mg/kg) or pentobarbital (3–6 mg/kg) were used, with continuous monitoring of the airway and pulse oximetry by a trained registered nurse.

Blinded review of SPECT images

An initial visual analysis of periictal scans for localization of a region of the highest increase in the ictal perfusion was performed by a nuclear medicine expert blinded to all patient data. A gray-scale of the images was then modified to achieve the best localization of the region with increased perfusion. For purposes of study analyses, SPECT images were independently reevaluated by three reviewers (PK, BM, AJ) who were aware of the pathologic diagnosis but not informed about clinical, EEG, or MRI data or postoperative seizure outcomes. Differences between the reviewers led to a case being re-reviewed together until a consensus was reached.

In the first step of the images' analysis, the reviewers were required to localize cortical hyperperfusion zones to 11 anatomically defined regions (frontal central, mesial, convexity, polar, basal; temporal mesial and lateral; parietal mesial and lateral; occipital mesial and lateral) or to classify the images as “nonlocalizing.” The extent of cortical hyperperfusion was then classified as follows: (1) focal (confined to a region of one gyrus or two contiguous gyri), (2) lobar, (3) multilobar, (4) hemispheric, and (5) bilateral. Subcortical activation of the cerebellum, basal ganglia, and thalamus was also assessed and classified as right, left, bilateral, and absent.

Completeness of hyperperfusion zone removal

In the next step of the analyses, both SPECT and postsurgical MRI images were presented to the reviewers in order to correlate SPECT findings with the resection site. If a localized cortical SPECT hyperperfusion zone was present, its relation to the resection cavity at the postsurgical MRI was classified as: (1) completely resected, (2) partially resected, and (3) nonresected (Fig. 1). In addition, subcortical activations were also related to the resection site and classified as (1) ipsilateral, (2) contralateral, and (3) bilateral.

Figure 1.

Examples of SPECT hyperperfusion zone and its relation to the resection cavity at the postsurgical MRI: (A) completely resected, (B) partially resected, (C) nonresected.

Relation of SPECT findings to EEG and MRI localizations

Finally, completeness of the SPECT hyperperfusion zone removal was correlated with completeness of resection defined by MRI and EEG findings. Completeness of resection is always determined at time of surgery by an epilepsy team including neurologists, neuroradiologists, and neurosurgeons. Complete resection is defined as the entire removal of the region of the MRI abnormality (if present) and the cortical region exhibiting prominent ictal and interictal abnormalities on intracranial EEG (Jayakar et al., 1994; Paolicchi et al., 2000; Krsek et al., 2009). Because SPECT findings were never considered when completeness of the resection was assessed, these two parameters (completeness of the SPECT hyperperfusion zone removal and completeness of the resection defined by MRI and EEG) could be compared as independent variables.

Following the comparison of completeness of SPECT hyperperfusion zone and the MRI/EEG-defined epileptogenic region with the anatomic limits of the excavation cavity, we identified four outcomes: (1) both the SPECT hyperperfusion zone and MRI/EEG-defined epileptogenic region were completely removed; (2) only the SPECT hyperperfusion zone was removed; (3) only the MRI/EEG-defined epileptogenic zone was removed; (4) resections of both regions were incomplete.

Surgical outcome

We analyzed postoperative seizure status 2 years after the final surgery as assessed during outpatient visits and telephone contact. Surgical outcome was classified according to Engel's classification scheme: (I) completely seizure-free, auras only or only atypical early postoperative seizures, (II) ≥90% seizure reduction or nocturnal seizures only, (III) ≥50% seizure reduction, and (IV) < 50% seizure reduction. For purposes of the study, seizure outcomes were classified as “favorable” in Engel I and II groups and “unfavorable” in Engel III and IV groups.

Statistical analysis

Completeness of the hyperperfusion zone removal as well as relation of SPECT hyperperfusion zone and MRI/EEG-defined epileptogenic region removals were statistically compared with postsurgical seizure outcome. The values of these variables were categorical, allowing the relationships among parameters to be evaluated with cross-tabulation tables by Pearson chi-square tests. Statistical evaluation was performed in three sets: (1) all patients, (2) previously nonoperated patients, and (3) patients after a failed epilepsy surgery evaluated before a reoperation. All statistical calculations were performed in STATISTICA (StatSoft, Inc., Tulsa, OK, U.S.A.) software.

Results

Patients

One hundred six subjects (49 male and 57 female) from a population of 567 patients operated between March 1986 and June 2006 met the above-mentioned criteria. According to Palmini and Luders classification (Palmini et al., 2004), there were 13 subjects with mild malformations of cortical development type II (mMCDs), 32 with FCD type Ia, 20 with FCD type Ib, 22 with FCD type IIa, and 19 with FCD type IIb. Mean age at surgery was 9.17 years (range 2 months–30 years). Forty-four patients underwent one-stage excisional procedures guided by preexcision electrocorticography; chronic invasive monitoring utilizing implanted subdural electrodes was performed in 62 subjects. Twenty-five patients underwent reoperation; three subjects underwent three surgeries.

Ictal SPECT studies

A total of 173 ictal SPECT studies were performed in subjects included in the study; there was a maximum of 6 SPECT studies in one subject. Mean age at SPECT study was 8.95 years (range 54 days–29 years). One hundred sixteen ictal SPECT studies were performed in nonoperated patients; 40 postsurgical ictal SPECT studies were done in subjects who subsequently underwent reoperation; 17 postsurgical ictal SPECT studies were performed as a part of the diagnostic work-up in patients who had no further surgery (these studies were excluded from the outcome analysis). Five ictal SPECT studies in three patients were done after a second epilepsy surgery.

Characteristics of ictal SPECT findings

Cortical hyperperfusion was present in 124 ictal SPECT studies (72%) and absent in 49 studies (28%). A localized hyperperfusion zone was found in 32 (80%) of 40 subjects evaluated after a failed excisional procedure. The extent of cortical hyperperfusion was classified as focal in 27 SPECT studies (22%), lobar in 45 studies (36%), multilobar in 40 studies (32%), hemispheric in 9 studies (7%), and bilateral in 3 studies (2%). Cerebellar activation was present in 148 ictal SPECT studies (86%), and activation of the basal ganglia and thalamus was observed in 127 studies (73%).

Characteristics of subcortical activations

In 124 SPECT studies with cortical hyperperfusion, cerebellar activation was contralateral to the cortical hyperemia in 62 studies (50%), ipsilateral in 4 studies (3%), bilateral in 44 studies (35%), and absent in 14 studies (11%) (Table 1). In 49 subjects without cortical hyperperfusion, cerebellar activation was unilateral in 10 studies (21%), bilateral in 28 studies (57%), and absent in 11 studies (22%).

Table 1. Characteristics of subcortical activation in 173 ictal SPECT studies performed in 106 subjects
 Cerebellar activation (%)Basal ganglia/thalamic activation (%)
Cortical hyperperfusion present (n = 124)
Contralateral62 (50)2 (1)
Ipsilateral4 (3)27 (22)
Bilateral44 (35)74 (60)
None14 (11)21 (17)
Absent cortical hyperperfusion (n = 49)
Unilateral10 (21)2 (4)
Bilateral28 (57)22 (45)
None11 (22)25 (51)

Basal ganglia/thalamic activations were analyzed in 124 SPECT studies with a cortical hyperperfusion; they were ipsilateral to the cortical hyperemia in 27 studies (22%), contralateral in two studies (1%), bilateral in 74 studies (60%), and absent in 21 studies (17%). In 49 studies without cortical hyperperfusion, basal ganglia/thalamic activations were unilateral in only two studies (4%), bilateral in 22 studies (45%), and absent in 25 studies (51%).

Completeness of the hyperperfusion zone removal and its relation to surgical outcome

Outcome analyses included 156 ictal SPECT studies performed in 106 subjects either before the first surgery or prior to reoperation (Table 2). Twenty-four patients had nonlocalized studies only; 82 patients had at least one localized SPECT study. In the latter group, the hyperperfusion zone was completely removed in 35, partially resected in 36, and nonresected in 11 patients. Favorable postsurgical seizure outcome was achieved in 16 patients (67%) with nonlocalized ictal SPECT, 30 patients (86%) with completely resected ictal hyperperfusion, 13 patients (36%) with partially resected ictal hyperperfusion, and 5 patients (45%) with nonresected ictal hyperperfusion (p = 0.000198). Similar results were found when nonoperated patients (N = 78) were analyzed separately (p = 0.000720); however, there was no statistically significant difference in studies performed after a previous brain resection (N = 28) (p = 0.193550).

Table 2. Relationship of completeness of the SPECT hyperperfusion zone removal to surgical outcome in patients with known postsurgical seizure outcome, focal or lobar SPECT studies, and normal MRI findings
 Favorable outcome (%)Unfavorable outcome (%)
All patients (n = 106, p = 0.000198)
No activation (n = 24)16 (67)8 (33)
Completely resected (n = 35)30 (86)5 (14)
Partially resected (n = 36)13 (36)23 (64)
Nonresected (n = 11)5 (45)6 (55)
Patients with focal or lobar SPECT studies only (n = 49, p = 0.007139)
Completely resected (n = 26)22 (85)4 (15)
Partially resected (n = 18)7 (39)11 (61)
Nonresected (n = 5)3 (60)2 (40)
Patients with normal MRI findings (n = 34, p = 0.054193)
No activation (n = 12)6 (50)6 (50)
Completely resected (n = 6)6 (100)0
Partially resected (n = 14)7 (50)7 (50)
Nonresected (n = 2)02 (100)

Patients with localized (focal and lobar) ictal SPECT studies and subjects with normal MRI findings were analyzed separately. In 49 children with localized ictal SPECT, favorable postsurgical outcomes were achieved in 22 (85%) subjects after complete resection, 7 (39%) with partial resections and 3 (60%) with nonresected hyperperfusion zones (p = 0.007139).

In a subset of 34 patients with nonlocalizing MRI findings, all 6 patients with completely resected zones of hyperperfusion had favorable postsurgical outcomes. In contrast, both patients with nonresected hyperperfusion zones had unfavorable outcomes. Seven (50%) of 14 patients with partially resected hyperperfusion zones had favorable postsurgical outcomes (p = 0.054193).

No statistically significant difference in predicting surgical outcomes was found between groups of patients with normal MRI findings and subjects with FCD type IIb (p = 0.03306 and p = 0.03759, respectively; four MRI-negative FCD type IIb subjects were excluded from this analysis). Complete resections of the hyperperfusion zone were more frequently encountered in FCD type IIb (10/15 subjects) than in MRI-negative patients (6/30 subjects).

Relation of SPECT findings and MRI/EEG localizations to surgical outcome

One hundred fourteen localized ictal SPECT studies performed in 82 patients were included in the analyses (Table 3). Both SPECT hyperperfusion zones and MRI/EEG-defined epileptogenic regions were completely removed in 30 patients. In five patients only the SPECT hyperperfusion zone was completely removed. Only the MRI/EEG-defined epileptogenic region and not the SPECT hyperperfusion zone was removed in 22 patients. Resections of both regions were incomplete in 25 subjects.

Table 3. Relationship of surgical outcome to complete removal of the SPECT hyperperfusion zone and MRI/EEG-defined epileptogenic region (ER)
 Favorable outcome (%)Unfavorable outcome (%)
Completeness of removal of individual regions related to seizure outcome (n = 82, p = 0.000009)
Removal of both regions complete (n = 30)26 (87)4 (13)
Only SPECT hyperperfusion removed (n = 5)4 (80)1 (20)
Only MRI/EEG-defined ER removed (n = 22)13 (59)9 (41)
Removal of both regions incomplete (n = 25)5 (20)20 (80)
SPECT hyperperfusion removed (n = 35)30 (86)5 (14)
MRI/EEG-defined ER removed (n = 52)39 (75)13 (25)

Favorable postsurgical outcome was achieved in 26 subjects (87%) after complete removal of both regions compared to 5 patients (20%) in whom removal was incomplete. After the SPECT hyperperfusion zone but not the MRI/EEG-defined epileptogenic region was removed, 4 patients (80%) had a favorable postsurgical outcome. Conversely, the complete removal of the MRI/EEG-defined epileptogenic region and not the SPECT hyperperfusion zone achieved favorable outcomes in 13 subjects (59%). Differences between individual groups of patients were highly statistically significant in the cohort of all patients (N = 82, p < 0.000009) and nonoperated patients (N = 36, p = 0.007145), but there was no statistically significant difference in subjects after reoperation (N = 25, p = 0.058040).

Discussion

The present study analyzed the largest population of pediatric patients with cortical malformations undergoing ictal SPECT before excisional epilepsy surgery. The size of this data set of patients with cortical malformations greatly facilitates the ability to define its practical value for surgical management.

We showed that ictal SPECT is a highly effective localizing tool in the pediatric mMCD/FCD population. Seventy-two percent of SPECT studies revealed localized cortical hyperperfusion. This level of efficacy is comparable to that of previous studies in adults with extratemporal focal epilepsy (66.7%, O'Brien et al., 2000) and FCD (86%, O'Brien et al., 2004) and superior to the previous reported rate of 53% in pediatric FCD (Gupta et al., 2004). Furthermore, the extent of hyperperfusion was focal or lobar in the majority of SPECT studies (58%), whereas multilobar activations occurred in only 32% of studies and hemispheric or bilateral findings were rare (7% and 2%, respectively). The number of multilobar activations is consistent with the significant occurrence of extensive cortical malformations in children as we have previously noted (Krsek et al., 2008, 2009).

Complete resection of the region of ictal SPECT hyperperfusion strongly predicted favorable postsurgical seizure outcomes in our cohort. Eighty-four percent of subjects with a completely resected zone of ictal hyperperfusion had a good result. Ictal SPECT thus provided important additive data to the localization rendered by EEG and MRI. Although we recognize that ictal SPECT activations may represent propagated sites that do not require resection, our study results suggest that ictal activations when localized, may be used to guide resections beyond the anatomic or electrographic abnormalities to achieve better outcomes. Pending prospective confirmation, ictal SPECT may play a significant role in surgical planning.

In a previous pediatric FCD series, 85% of seizure-free patients evidenced ictal SPECT findings that were concordant with the resection site (Gupta et al., 2004). In a series of adults with extratemporal focal epilepsy (O'Brien et al., 2000), an excellent outcome was achieved in 58% of subjects with localizing SISCOM findings concordant with the surgical site. In a heterogeneous pediatric surgical population (O'Brien et al., 1998), 85.7% of patients with SPECT findings concordant with the resection site achieved a superior outcome. Kaminska et al. (2003) reported that the maximal cerebral blood flow changes between ictal and interictal SPECT colocalized with the resection site in 70% of patients with favorable surgical outcome; the difference between subjects with nonlocalizing or nonconcordant ictal SPECT findings was also significant.

Seizure outcome in subjects with partially resected SPECT hyperperfusion zones (36% with favorable outcome) was comparable to that of patients with nonresected ictal hyperperfusion (45% with favorable outcome). This finding clearly shows that only complete removal of the hyperperfusion zone predicts surgical success. It should be noted, however, that different patterns of hyperperfusion are observed in patients with FCD. Using the SISCOM method, Dupont et al. (2006) described four distinct hyperperfusion patterns associated with focal dysplastic lesions, representing different modes of seizure propagation. Complete resection of the SISCOM hyperperfusion cluster was not required for seizure freedom. In accordance with these findings, a subset of our patients also had a favorable seizure outcome following incomplete removal of the hyperperfusion zone. This is an important observation as complete resection of the hyperperfusion zone is not always possible, for example, because of overlap with eloquent cortical areas.

Ictal SPECT in our series had the same localizing value in nonoperated and operated patients. The majority of subjects (17 of 28) being evaluated for repeat resection had a localized hyperperfusion zone. Wetjen et al. (2006) reported an even a higher percentage (79%) of localized hyperperfusion zones in 58 previously operated patients evaluated by SISCOM. In our series and Wetjen's cohort, complete resection of the hyperperfusion zone reliably predicted a favorable seizure outcome after reoperation. We therefore believe that ictal SPECT is a powerful diagnostic tool for patients who have failed epilepsy surgery.

Our study design allowed for a direct comparison of MRI- and EEG-defined completeness of resection with complete removal of the SPECT hyperperfusion zone. At our institution, completeness of resection is always determined on the basis of the postoperative MRI and intracranial EEG data (Jayakar et al., 1994; Paolicchi et al., 2000; Krsek et al., 2009). SPECT data are not factored into this determination, although ictal SPECT findings are part of the presurgical workup with the potential to influence the site and extent of resection. However, the evaluation of the completeness of the SPECT hyperperfusion zone removal was performed retrospectively by experts not involved in the initial assessment of the completeness of resections who were blinded to patient data.

The study results reveal that complete resection of the region of ictal SPECT hyperperfusion was an equally important predictor of favorable surgical outcome when compared with the removal of the epileptogenic region defined by MRI and EEG. Favorable postsurgical outcome was achieved in 75% of patients with complete removal of the epileptogenic region defined by MRI and EEG, and in 86% of subjects with complete resection of the SPECT hyperperfusion zone. By analyzing all pediatric mMCD/FCD patients operated on in Miami Children's Hospital, we previously showed that 83% of patients undergoing complete resections based on combined MRI/EEG criteria achieved favorable postsurgical seizure outcomes (Engel class I + II categories) compared to only 28% after incomplete excision of the dysplastic region (Krsek et al., 2009). The lower success rate in the current patient group could be explained by a selection bias, in that ictal SPECT is rarely employed in straightforward lesional cases that already have a high probability of postsurgical seizure freedom.

Our results suggest that the addition of ictal SPECT complements the utility of MRI and EEG. However, a limitation of our study is that it was retrospectively based on ictal SPECT data that confirmed the general location of the epileptogenic zone rather than delineating it. Based on the present data we could not quantify how much SPECT improved surgical outcomes or precluded the need for invasive investigation. Our favorable experience with SPECT nevertheless suggests that a proportion of FCD cases may benefit from one-stage multimodality image-guided surgeries employing the SISCOM method to improve seizure outcome.

Favorable postsurgical outcomes were more frequently encountered after complete resection of the SPECT hyperperfusion zone than after complete removal of the MRI/EEG-defined epileptogenic region. However, this observation should be approached with caution as the extent of SPECT hyperperfusion is often larger than MRI- or intracranial EEG-defined abnormities and therefore would be expected to lead to more extensive resections. Larger resections would logically be expected to have a higher likelihood of seizure freedom and a higher rate of complication. Because hyperperfusion patterns also reflect zones of seizure propagation, complete resection of the hyperperfusion zone may not always be necessary.

We specifically analyzed subsets of patients with localized ictal SPECT studies and normal MRI findings. When the SPECT hyperperfusion zone was completely removed, favorable postsurgical outcomes were achieved in 85% of subjects with localized SPECT findings and in all patients with nonlocalizing MRI studies. This outcome is in accord with previous observations showing accurate seizure localization of ictal SPECT in nonlesional epilepsy cases (O'Brien et al., 2000, 2004; Cascino et al., 2004). We, however, found the same localization yield in subjects with FCD type IIb, suggesting that ictal SPECT is equally valuable in both patient populations. In contrast, the localizing value of ictal SPECT is limited in subjects with large (multilobar and hemispheric) hyperperfusion zones.

In accord with a majority of previously published studies, contralateral cerebellar activations (62 of 66 studies with an asymmetric activation) and ipsilateral basal ganglia/thalamic activations (27 of 28 asymmetric studies) predominated in SPECT studies when cortical hyperperfusion was present. In a series of patients with temporal and extratemporal epilepsy, 75% of asymmetric cerebellar activations were contralateral to the seizure focus defined by electroclinical, ictal SPECT, and MRI data (Won et al., 1999). Sojkova et al. (2003) found that 77% of basal ganglia and 80% of thalamic activations were ipsilateral to the seizure focus defined by electroclinical data and surgical cure. Other studies analyzing subcortical activation focused mainly on temporal lobe epilepsy. Shin et al. (2001) described basal ganglia activation ipsilateral to the side of the epileptogenic temporal lobe in 13 of 17 studies with unilateral basal ganglia activation and contralateral cerebellar activation in 15 of 25 studies with unilateral cerebellar activation. However, Van Paesschen et al. (2003) did not find basal ganglia activation in SPECT studies performed during complex partial seizures in patients with hippocampal sclerosis, and surprisingly found contralateral cerebellar hypoperfusion in 100% and ipsilateral cerebellar hyperperfusion in 87.5% of studied patients.

We also sought to ascertain whether subcortical activation has lateralizing value in SPECT studies lacking cortical hyperperfusion. Cerebellar and basal ganglia and thalamic activation did not lateralize the epileptogenic zone, as subcortical activation was mostly bilateral (57% of cerebellar and 45% of basal ganglia/thalamic activations) or absent (22% of cerebellar and 51% of basal ganglia/thalamic activations).

Finally, we wish to emphasize that visual evaluation of SPECT data and not the SISCOM technique is employed at our institution. Several studies found SISCOM superior to the visual inspection of SPECT images in the seizure focus localization. O'Brien et al. (1998) reported better interrater agreement for two independent reviewers for SISCOM compared to visual interpretation of SPECT data (84.3% vs. 41.2%). Lack of interictal SPECT may lead to both false-negative and false-positive results (Lewis et al., 2000). Moreover, postictal suppression of perfusion in epileptogenic region could have occurred with brief seizures (O'Brien et al., 1999). A multicenter prospective study (Matsuda et al., 2009) found that SISCOM had a higher predictive value of good surgical outcome and more reliability for defining the epileptogenic focus than side-by-side inspection of images. Therefore, we recognize that SISCOM could better delineate the epileptogenic zone than ictal hyperperfusion that reflects propagation sites as well. However, our results are in all aspects comparable to studies using SISCOM technique. Moreover, SISCOM has added costs (ictal and interictal studies are both necessary and children frequently require sedation). We therefore suggest that visual interpretation of SPECT data might be sufficient for planning epilepsy surgery in children.

Acknowledgments

Supported by grants Kontakt Program ME09042, GAUK 17010, and CZ.2.16/3.1.00/24022, and by the project for conceptual development of research organization 00064203.

Disclosure

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.

Ancillary