Among all cerebral regions, the orbitofrontal (OF) cortex is one of the least studied and understood region. It has vast bidirectional connectivity to a widely distributed network involving the frontal lobe, the temporal lobe, and the limbic system, making localization of ictal foci to this region particularly difficult (Cavada & Schultz, 2000; Alexopoulos & Tandon, 2008). Because of the hidden, deep location of the OF region in relation to the scalp electrode sampling, scalp electroencephalography (EEG) can be at times misleading with the lack of distinct ictal manifestations (Ebner, 2001; Pedley et al., 2003; Burgess et al., 2005).
High-resolution structural magnetic resonance imaging (MRI), when interpreted by experienced neuroradiologists, can be helpful to search for focal abnormalities in the OF region. When a patient presents with negative preoperative MRI, imaging with higher magnetic field or three-dimensional volume sequence, postprocessing analysis, as well as functional neuroimaging such as positron emission tomography (PET), single-photon emission computed tomography (SPECT) and magnetoencephalography, may be important complements to the test battery of presurgical evaluation (Knowlton et al., 2008; Bien et al., 2009). In the literature, there are only rare cases documenting clear OF epilepsy with a negative preoperative MRI (Rugg-Gunn et al., 2002; Smith et al., 2004; Kriegel et al., 2012). Moreover, there was no study specifically reporting postprocessing neuroimaging characteristics of OF epilepsies in the setting of a normal preoperative MRI.
Recently, a voxel-based MRI postprocessing technique has been used to detect subtle cortical malformations missed with conventional MRI visual inspection (Huppertz et al., 2005, 2008; Wagner et al., 2011). This technique compares each individual patient with a normal control database, and yields three-dimensional (3D) feature maps of gray–white junction, cortical gyration, and cortical thickness. Abnormalities in these feature maps can indicate underlying focal cortical dysplasia (FCD; Krsek et al., 2008). The yield and specificity of the findings using this postprocessing technique in patients with OF epilepsy have not been reported previously.
In this study, we propose to examine the image postprocessing and functional neuroimaging (18F-fluorodeoxyglucose [FDG]-PET and ictal SPECT) characteristics of a strictly defined and highly selected, invasive EEG-confirmed group of patients with pharmacoresistant OF epilepsy with negative preoperative MRI. All patients were rendered seizure-free after resection of the OF region with/without adjacent cortex. In addition, we report the electroclinical features obtained from noninvasive and invasive studies and pathologic findings in these patients.
- Top of page
- Supporting Information
We present here the largest series to date of MRI-negative orbitofrontal epilepsy. Our cohort of OF epilepsy was confirmed by invasive EEG recordings and postoperative seizure freedom. Our data show that voxel-based morphometric MRI postprocessing can identify subtle abnormalities in the majority of patients we studied. Our results suggest that a MAP+ abnormality may guide a more focused invasive EEG strategy and a subsequent tailored surgical resection.
Despite the significant improvements in MRI technology, a significant number of patients with frontal lobe (and in particular orbitofrontal) epilepsy continue to have “normal” MRI studies upon visual analysis (Lorenzo et al., 1995; Noe et al., 2013; See et al., 2013). The difficulty in identifying lesions within the OF region in particular may be due to the following possibilities: (1) the sulcation pattern in the OF region is highly variable and densely packed; (2) some MRI sequences (in particular, T2-weighted sequences) have higher susceptibility artifacts/geometric distortion due to the proximity of the OF region to the air-filled sinuses; therefore, they are not providing adequate anatomic information at times (Kringelbach & Rolls, 2004). For these reasons, subtle lesions in the OF region may not be obvious upon conventional MRI visual inspection. When noninvasive evaluation data (e.g., scalp EEG) do not point to a specific area of interest, MRI readers lack a testable anatomic hypothesis, and subsequently the study may be read as negative. Under these circumstances, a whole-brain MRI postprocessing technique that directs the reader's attention to suspicious abnormalities may prove to be helpful.
Focal cortical dysplasia, a common substrate in patients with refractory epilepsies, is frequently associated with blurring in gray–white junction (Krsek et al., 2008). Therefore, it is not surprising to find abnormalities on MAP junction file to be associated with FCD histopathology in two patients (P1 and P5). Of interest, histopathology findings also showed two MAP+ patients had nonspecific gliosis, and the MAP-negative patient had FCD in the resected specimen. One explanation could be the limited sampling issue of the pathologic tissue undergoing examination, under which circumstances single/isolated/small lesions could be missed. Another possible explanation is the hypothesis that MAP, as a postprocessing technique of structural MRI, is not sensitive strictly to FCD per se, but to any pathologic substrate (or combinations of them) causing T1 signal alteration leading to a blurred gray–white junction. This hypothesis is also consistent with the non-FCD pathology finding in P4. Consistent with our study, a 2004 study also reported one patient with “nonlesional” OF epilepsy whose surgical pathology showed only gliosis; the patient remained seizure-free at 5-year follow-up (Smith et al., 2004). A 2002 case study presented another patient without abnormality on conventional MRI but with increased diffusivity in the OF region. The patient underwent inferior frontal lobe resection and remained Engel class IIa. Pathology also showed gliosis but not dysplasia (Rugg-Gunn et al., 2002).
Although most identified MAP+ abnormalities were correlated with in situ epileptogenicity, our data show that the presence of MAP+ abnormalities does not always correlate directly with active epileptogenicity. These abnormalities were not resected and no information regarding the histologic characteristics can be obtained. As observed in P5, the resection of the structurally abnormal and EEG-proven epileptic abnormality leads to long-lasting seizure control (>5 years). The finding of structural abnormalities in areas outside the presumed epileptic focus is consistent with the literature (Colliot et al., 2006; Salmenpera et al., 2007; Fauser et al., 2009; Yasuda et al., 2010). These postprocessing imaging changes could be due to potentially epileptic/proepileptic abnormalities that were not epileptic at the time of invasive recordings, and may be the underlying cause of late seizure recurrence following epilepsy surgery (Najm et al., 2013). Another explanation of the presence of “nonepileptic” focal abnormalities on MAP is the possibility of false-positive changes. Previously published automated voxel-based morphometry studies had reported false-positive findings in the control group, particularly when the sensitivity of the patient group was maximized (Focke et al., 2008). The significance of MAP+ regions should therefore be interpreted in the context of all other clinical test results.
Interictal and ictal scalp EEG rarely provide localizing information in OF epilepsy due to the large distance between the epileptogenic zone and the electrodes (Alexopoulos & Tandon, 2008); however, they can still have lateralizing value, as suggested by our data. Similar to several previous studies, we found interictal and ictal EEG often falsely localized to the ipsilateral temporal region, although it is interesting to note that bifrontal discharges and onset were not seen in our group (Ludwig et al., 1975; Chang et al., 1991; Jobst & Williamson, 2005). False localization to the temporal region occurs probably due to the close bidirectional connections between the temporal and OF region, as demonstrated by a classic study in 1958 using strychnine neuronography, in which the authors found that spikes generated in the OF region can quickly propagate to the ipsilateral temporal cortex and vice versa (Kendrick & Gibbs, 1958).
Intracranial EEG remains the corner stone of orbitofrontal epilepsy evaluation. It is especially illustrative once a reasonable implantation hypothesis has been made based on the noninvasive evaluation. Intracranial EEG studies in all our patients sufficiently covered the OF region and exclusively pointed ictal onset to the OF region in all patients. The basal frontal region is usually sampled by a four-by-four subdural electrode array (shown in Fig. 1, P2, P3 and Fig. 2, P5), whereas stereotactic EEG can be especially useful to explore the generators located in the mesial and lateral aspects of the OF region (Fig. 1, P1, P4, and Fig. 3, P6; Munari & Bancaud, 1992). Due to the widespread connection from the OF area, coverage of the anterior and mesial temporal regions, insula, operculum, and cingulate gyrus is recommended to improve the investigators' ability to differentiate seizure onset and propagation (Alexopoulos & Tandon, 2008).
Although FDG-PET has some diagnostic value in nonlesional frontal lobe epilepsies as a whole (Lee et al., 2005; Salamon et al., 2008), there is no previous literature specifically studying its effectiveness in OF epilepsies. In all patients included in this study, PET was negative, multilobar, or falsely localizing to the temporal lobe. Only on retrospective review can one find concordance between hypometabolism on PET and the MAP+ foci/intracranial EEG onset. Several reasons could contribute to the relative low yield of PET to OF epilepsies, including complicated gyration pattern in the OF region, which may lead to partial volume effect smearing the image resolution, and the fact that PET was visually inspected after coregistration with the MRI, without further quantitative analysis. Alternatively, perhaps epileptic activities originating from the OF region just spreads more quickly. As illustrated in previous studies, fast propagation of epileptic activities can be associated with widespread hypometabolism, or hypometabolism remote to the ictal-onset zone (Wong et al., 2010, 2012).
No previous studies specifically examined the effectiveness of SPECT SISCOM in localizing seizures from the OF region. In our cohort, SPECT was recorded in four of the six patients, but none localized exclusively to the OF region despite early injection. The hyperperfusion always existed in multiple adjacent areas. The low efficacy of SPECT in this cohort can be explained by rapid propagation of seizures to regions closely connected with OF area (Noachtar et al., 1998).
The gold standard for accurate localization of the epileptogenic zone is invasive EEG recording over the seizure-onset zone, with postresective seizure freedom (Engel et al., 1981; Rosenow & Luders, 2001; Kellinghaus & Luders, 2004). Patients with OF epilepsy carrying the strongest proof should not only have ictal onset exclusively from the OF region, but should also be rendered seizure-free after exclusive resection of the OF cortex. However in practice, the OF cortex is rarely the only area of resection. Overall, multiple concordant modalities generally lead to a more confined resection. It is our hope that this study, with the addition of MRI postprocessing findings, could shed light on future refinement of surgical strategies of orbitofrontal epilepsies.
Another potential benefit of our study is the contribution of MRI postprocessing in the identification of a possibly epileptogenic lesion in an area of the brain that is rarely suspected as an anatomic location of frontal lobe epilepsy and frequently mislocalized to other neighboring regions. This hypothesis can be tested through MAP analyses of patients who failed previous frontal or temporal lobe resections, and verification of seizure-free status with subsequent resection of MAP+ areas (e.g., P1).
In summary, our study highlights the importance of MRI postprocessing as a promising tool to complement conventional visual inspection of the MRI in terms of identifying potentially epileptogenic but subtle cortical abnormalities. This tool, when used in conjunction with other noninvasive modalities, has the potential to assist the planning of invasive electrode implantation and surgical resection in patients with MRI-negative pharmacoresistant orbitofrontal lobe epilepsy.