The role of hybrid FDG-PET/MRI on decision-making in presurgical evaluation of drug-resistant epilepsy

Background When MRI fails to detect a potentially epileptogenic lesion, the chance of a favorable outcome after epilepsy surgery becomes significantly lower (from 60 to 90% to 20–65%). Hybrid FDG-PET/MRI may provide additional information for identifying the epileptogenic zone. We aimed to investigate the possible effect of the introduction of hybrid FDG-PET/MRI into the algorithm of the decision-making in both lesional and non-lesional drug-resistant epileptic patients. Methods In a prospective study of patients suffering from drug-resistant focal epilepsy, 30 nonlesional and 30 lesional cases with discordant presurgical results were evaluated using hybrid FDG-PET/MRI. Results The hybrid imaging revealed morphological lesion in 18 patients and glucose hypometabolism in 29 patients within the nonlesional group. In the MRI positive group, 4 patients were found to be nonlesional, and in 9 patients at least one more epileptogenic lesion was discovered, while in another 17 cases the original lesion was confirmed by means of hybrid FDG-PET/MRI. As to the therapeutic decision-making, these results helped to indicate resective surgery instead of intracranial EEG (iEEG) monitoring in 2 cases, to avoid any further invasive diagnostic procedures in 7 patients, and to refer 21 patients for iEEG in the nonlesional group. Hybrid FDG-PET/MRI has also significantly changed the original therapeutic plans in the lesional group. Prior to the hybrid imaging, a resective surgery was considered in 3 patients, and iEEG was planned in 27 patients. However, 3 patients became eligible for resective surgery, 6 patients proved to be inoperable instead of iEEG, and 18 cases remained candidates for iEEG due to the hybrid FDG-PET/MRI. Two patients remained candidates for resective surgery and one patient became not eligible for any further invasive intervention. Conclusions The results of hybrid FDG-PET/MRI significantly altered the original plans in 19 of 60 cases. The introduction of hybrid FDG-PET/MRI into the presurgical evaluation process had a potential modifying effect on clinical decision-making. Trial registration Trial registry: Scientific Research Ethics Committee of the Medical Research Council of Hungary. Trial registration number: 008899/2016/OTIG. Date of registration: 08 February 2016.

The aim of the present study was to investigate the possible effect of the results of hybrid FDG-PET/MRI on the decision-making by the epileptologist in both lesional and nonlesional drug-resistant epileptic patients. For this purpose, we selected two drug-resistant epileptic patient groups of the same size (either lesional or nonlesional cases), where the noninvasive video-EEG monitor and brain MRI were not conclusive regarding the resective surgery. We hypothesized that the results of hybrid FDG-PET/MRI may affect the initial judgement of the presurgical team: (1) the patient is eligible for iEEG exploration; or (2) eligible for resective surgery; or (3) the patient is not eligible for any further invasive procedure. Thus, in the present study, findings related to the alterations of decision-making were analyzed in detail with respect to imagery results (FDG-PET and MRI).

Subjects
We prospectively selected 60 adult patients (35 males, 25 females, mean age: 33.02, range:18-55 years) undergoing pre-surgical evaluation for drug-resistant, focalonset epilepsy at two tertiary academic medical centers: (1) Department of Neurology, University of Pécs and (2) National Institute of Clinical Neurosciences, Budapest. Of the patients, all suffering from focal-onset epilepsy, 52 were right-handed and 8 were left-handed. Patients with previous history of epilepsy surgery were not included. Patients were divided into groups based on whether they were lesional (30 patients, for details please see Table 2) with discordant investigational results or nonlesional (30 patients, for details please see Table 1).

Procedure and material
All of the patients underwent presurgical examinations: routine epilepsy clinic visits, noninvasive video-EEG monitoring, cranial MRI following the standard epilepsy protocol [25][26][27][28] and clinical semiology was evaluated by two presurgical teams. Before FDG-PET/MRI became available, some patients underwent PET/CT when MRI was negative or clinical and EEG findings suggested multiple seizure foci. An algorithm describing the steps of presurgical evaluation at our centers can be seen on A hybrid FDG-PET/MRI system (Siemens Biograph mMR, Siemens Heathineers, Erlangen, Germany) consisting of 3 T Verio magnet and MR compatible LSO crystal based APD PET detector system allowing simultaneous PET/MRI acquisition was used. The device was settled in Baka József Diagnostic Center, Kaposi Mór Hospital, Kaposvár, Hungary.
The fluorine-18 fluoro-2-deoxyglucose ([18F] FDG) PET imaging was performed according to the guideline of European Association of Nuclear Medicine Neuroimaging Committee [29]. All patients fasted for at least 6 h before the scan and blood glucose levels were checked prior to FDG administration. Before the scanning procedure the patients were asked to empty their bladder and were positioned comfortably in a quiet, dark room equipped with a video camera. Then a cannula for intravenous administration was placed. A 2 h-long supervision of the patient and 30 min of video EEG recording was performed before the administration of FDG (bolus of 200 MBq i.v.). We maintained the video-EEG monitoring for the whole uptake phase of FDG to ensure interictal state. If any seizure activities were recorded during video-EEG monitoring, the FDG-PET/MRI investigation was postponed to the following day. The PET/ MRI scan was started 60 min after the FDG administration. For the prevention of movement artifacts, we informed the patients that they should avoid voluntary movements in the scanner. After the scanning procedure the patients were further supervised for two half-lives of the radioisotope decay (ca. 240 min). To ensure complete simultaneous PET coverage, a 20 min and a 35 min list mode PET acquisition were applied. For PET attenuation correction purposes vendor-provided T2 ultrashort echo time (UTE) sequence was used, μMaps were MRI (both the prior studies and those with the accompanying PET) and PET studies were downloaded and blindly and separately re-interpreted, by two neuroradiologists (for MRI: PB and ZV) and nuclear medicine physicians (for PET: KB and ZT). The two MR studies were interpreted separately from each other by each of the neuroradiologists (PB, ZV). Finally, simultaneously acquired MRI and PET images were evaluated on fused images, and clinical decisions were done with these data.
New variables: MR status change, additional information provided by PET comparing to MRI and clinical decision according to PET/MRI results

MR status change
We created five new categories to describe the changes: nn: The patient was nonlesional prior to the study and he/she also remained nonlesional in this study. np: The patient was nonlesional prior to the study and changed to lesional in this study. nc: The patient was suspect for lesional prior to the study but the lesion was not confirmed in this study. pp.: The patient was lesional prior to the study and the study confirmed the original lesion. pp.+: The patient was lesional prior to the study and the study both confirmed the original lesion and found new epileptogenic lesion(s).
Additional information provided by PET comparing to MRI (indicating the localisation of glucose hypometabolism in nonlesional case as well as in lesional case, related to the MRI-lesion) A. Positive, revealing unilateral area(s) of hypometabolism in a nonlesional case.
B. Positive, revealing bilateral area(s) of hypometabolism in a nonlesional case.
C. Positive, ipsilateral, related to the MRI-identified lesion in a lesional case.
D. Positive, ipsilateral, but not related to the MRIidentified lesion, pointing to new area(s) within the same hemisphere in a lesional case.
E. Positive, contralateral to MRI-lesion in a lesional case.
F. Positive, bilateral in lesional case in a lesional case. G. Negative.

Clinical decisions according to PET/MRI results
The clinical decisions made by a consensus of the two multidisciplinary epilepsy surgery teams at two tertiary academic medical centers were classified into six categories:

Statistical analysis
We applied the Chi-square independence test to determine if there is a significant relationship between "MRI status change" and "Additional information provided by PET comparing to MRI" and used Pearson residuals to highlight the most contributing between these category pairs. Because of the small sample size, we performed Fisher's exact test to confirm the result of the Chisquare test.

Nonlesional group
Hybrid PET/MRI examination revealed that 18 of 30 patients were found to have new specific epileptogenic MRI-lesion(s), while 29 of 30 patients had an abnormal FDG uptake. Two lesions were found in five patients. Hybrid PET/MRI helped to indicate resective surgery instead of iEEG monitoring in 2 patients due to congruent MRI, PET, semiological and EEG data; resective surgery was performed in both cases with Engel I outcome (for details, please see Table 1). In seven patients, hybrid PET/MRI enabled us to decide to avoid any further invasive diagnostic procedures due to multiple electroclinical and/or hypometabolic epileptic foci with MRI-negativity (6 patients) or negative imagery (both PET and MRI) results (1 patient). Of the remaining 21 patients referred to iEEG due to discordant electroclinical and PET/MRI results, in 11 cases one MRI-target was revealed; in 5 patients two potential epileptogenic lesions were found; while 5 cases remained nonlesional. In 3 of 11 patients of whom one MRI-target was revealed iEEG was performed concluding in resective surgery in 2 cases with Engel III and Engel IV outcome; the remaining case was considered as not operable and did not concede to perform DBS/VNS implantation.
Assessing the data comparing the structural highquality MRI from hybrid FDG PET/MRI ("MRI status change") with the information providing the combination of glucose hypometabolism with structural data ("Additional information provided by PET comparing to MRI"), the Chi-square independence test and Fisher's exact test showed a significant association (p < 10e-6 and p < 10e-10, respectively). The Pearson residuals suggested that the group "np" (the patients were nonlesional prior to the study and changed to lesional in this study) and group "A" (PET revealed unilateral area(s) of hypometabolism in nonlesional cases) have a strong positive association as well as between group "nn" (patients were nonlesional prior to the study and they also remained nonlesional in this study) and group "B" (PET revealed bilateral area(s) of hypometabolism in nonlesional cases).

Lesional group
Two patients became eligible for resective surgery, because of concordant electroclinical and hybrid FDG-PET/MRI results, which confirmed the MRI-lesion visualized prior to this study. In one patient any further invasive investigation was found to be contraindicated due to discordant electroclinical and PET data; but more importantly, because of altered MRI-status (nonlesional). iEEG was planned in 27 patients, of whom three patients became eligible for resective surgery because of concordant hybrid FDG-PET/MRI results and electroclinical data; resective surgery was performed with Engel I outcome in 1 case and Engel II outcome in 2 cases (for details, please see Table 2).
Six of the remaining 27 patients proved to be inoperable instead of iEEG, 4 of these patients were found not to be confirmed lesional. In 3 patients, any further invasive procedure was contraindicated. Eighteen patients remained candidates for iEEG due to discordant electroclinical and FDG-PET/MRI results. In 6 of 18 patients iEEG was already performed concluding in resective surgery in all cases with Engel I (3 cases) and Engel II (3 cases) outcome.
Statistical analysis of the data comparing the structural high-quality MRI from hybrid FDG PET/MRI with the information providing the combination of glucose hypometabolism with structural data did not reveal strong association in any case in lesional group.

Discussion
In the case of drug-resistant focal-onset epilepsy, the most important issue in the algorithm of the decisionmaking is to judge whether a drug-resistant epilepsy patient is eligible for (1) resective surgery, (2) iEEG monitoring, or, (3) not eligible for any further invasive procedures [7,8].
PET/MRI coregistration has long been utilized in epilepsy centers and has been useful to guide a second look at MRI previously reported as nonlesional, thus guiding decision-making [17]. In an earlier publication, PET/ MRI coregistration helped to find obvious lesion in 6 of 10 nonlesional drug-resistant epilepsy patients [18]. In a pediatric study, 31 consecutive pediatric nonlesional epilepsy patients were reported, of whom nine showed subtle pathologic abnormalities after second MRI-reading guided by PET/MRI coregistration [19]. In another study, 35 consecutive epilepsy patients with refractory focal epilepsy were investigated: structural MRI showed no lesion in 15 patients, of whom PET/MRI coregistration detected hypometabolism in 7 cases that was undetected on PET alone [20]. In a recent paper, 103 consecutive epileptic patients with FCD type 2 were reported, of whom 61 patients were lesional, while 42 cases were dubious or negative. The additional value of PET/MRI coregistration in these 42 patients was predominant, because MRI localized FCD type 2 in 35 of 42 patients [21].
Here we report the role of hybrid FDG-PET/MRI on the decision-making workflow. Thus, in the present study, findings related to the changes in possible decisions of the presurgical team were analyzed in detail with respect to separate imaging results (FDG-PET and MRI) in both lesional and nonlesional epilepsy patients.
Our main finding is that hybrid FDG-PET/MRI decidedly influenced the decision-making of the presurgical team significantly. Its cardinal effect was the     figure), a left-right propagation in between a 10-s interval (red arrows, right side of the figure), which was remote-controlled by a possible left orbitofrontal seizure onset zone. The patient did not allow neither a second iEEG intervention, nor VNS or DBS implantation  Table 1, group np, decision type 1.) A drug-resistant epileptic patient with the electroclinical features of a right frontotemporal epilepsy. a During video-EEG monitoring, her habitual hypermotor seizure with a right frontotemporal seizure activity was registered (red arrows). Originally, she was nonlesional and became lesional in this study. b Coronal FLAIR images: white arrows show possible focal cortical dysplasia in the right anterior cingulate cortex (upper row), in the medial cortex of the right straight gyrus (middle row), and the mildly increased signal intensity and blurred cortex-white matter interface in the right temporal lobe (lower row) can be seen. 18F-FDG PET and PET/MRI presented c and e a hypometabolism in the right mesiofrontal region (white arrowheads), d as well as in the right temporal lobe (white arroewheads). This patient remained as iEEG candidate; iEEG monitoring has not yet been realised increased sensitivity of brain MRI in 60% of nonlesional patients, which is a principal component in judging the chance for seizure-freedom (nonlesional cases: 20-65% vs. lesional cases: 60-90%) [10,11].
In an earlier pilot study using hybrid PET/MRI, 11 epileptic patients were investigated by FDG-PET/MRI without gross structural abnormalities that could interfere with image processing. Unfortunately, it is not clear from this publication, whether or not new structural lesions were found [23].
In another pilot hybrid FDG-PET/MRI study, of the twenty-nine patients assessed who underwent epilepsy surgery evaluation, in four cases new structural MR lesions were detected with the aid of FDG-PET findings, and one patient showed a new abnormal hypometabolism without any MRI abnormality. All new FDG-PET/MR lesions were clinically significant with concordant EEG and/or SPECT results as potential epileptic foci [24]. Recently, the same research group reported that hybrid FDG-PET/MR  Table 2, group pp., decision type 2.): A drug-resistant epileptic patient with the electroclinical features of a right frontal epilepsy. a Video-EEG monitoring revealed his habitual seizure, a right frontocentral seizure activity was seen, which rapidly became bilateral (marked with red arrows). Concordantly, cranial MRI showed a nodular heterotopia in the right inferior frontal gyrus. b Axial T2 (left), coronal FLAIR (middle) and coronal T1 MPR (right) images. The white arrow on the T2 image and the large black arrows on the FLAIR and T1 images show focal nodular subependymal grey matter heterotopia. The small black arrows on the coronal T1 MPR image (right) show probable migrational bands. c Exceptionally compared to the other cases, during 18F-FDG PET and PET/MRI, a circumscribed FDG accumulation reaching the intensity of cortical tracer uptake (and highly exceeding white matter uptake) can be observed, identically to the right periventricular heterotopia. In this case, resective surgery became available instead of iEEG. Because the patient was left-handed, fMRI and also Wada-test were performed and they proved that in this case, active Broca region is localized in the right hemisphere. Thus, resective surgery was performed in awake state and finally, only a partial resection was possible. After the resective surgery, patient had much shorter (1-3 s long) seizures identified new structural or functional lesions in 10 of 74 patients [22].
Our study aimed to clarify the possible role of hybrid FDG-PET/MRI on decision-making in drugresistant partial-onset epilepsy patients as well as to compare this effect with earlier PET/MRI coregistration studies.

Nonlesional group
In our study, the hybrid FDG-PET/MRI revealed a new morphological lesion in 18 patients and PET hypometabolism in 29 patients within the nonlesional group. In this patient cohort, PET revealed unilateral area(s) of hypometabolism. However, PET revealed bilateral area(s) of hypometabolism in those patients who were  Table 2, group pp.+, decision type 3.) A drug-resistant epileptic patient with the electroclinical features of a bitemporal lobe epilepsy. a Video-EEG monitoring. During her habitual seizure, a right frontotemporal seizure rapid activity was seen (marked with red arrows). Meanwhile, original cranial MRI (made before this study) showed an FCD along the left collateral sulcus. b Cranial MRI made in this study (coronal FLAIR images): horizontal arrows show the originally detected FCD along the left collateral sulcus while the oblique arrows show the newly observed FCD along the right collateral sulcus. The vertical arrows show the typical configuration of bilateral hippocampal malrotation, while c and d 18F-FDG PET and PET/MRI presented a hypometabolism in the right and left frontotemporal lobe, with a right predominance (red boxes). In summary, this patient was considered as not eligible for any further invasive procedures instead of iEEG nonlesional prior to the study and also remained nonlesional in this study. These two correlations were proved to be statistically significant.
Due to hybrid FDG-PET/MRI results, resective surgery was indicated instead of iEEG monitoring in 2 cases with an Engel I outcome; hybrid FDG-PET/MRI results helped to avoid any further invasive diagnostic procedures in 7 patients. The remaining 21 patients were referred to iEEG. In 16 of the remaining 21 patients, novel specific epileptogenic MRI-lesion(s) were revealed, proposing potential target(s) for iEEG monitoring, hopefully increasing the chance of successful identification of the epileptogenic zone. Three of them were underwent iEEG and 2 of them resective surgery with Engel III and IV outcome.
As we showed in Methods section, (beyond better image quality/resolution) a major difference was the application of 3D FLAIR sequence which was lacking in the earlier imaging protocol. This might in part explain the newly identified specific epileptogenic lesions in a significant proportion of the nonlesional group. Nine of these 18 patients showed the newly detected lesions exclusively on the 3D FLAIR images. Another possible explanation is the growing body of experience of our neuroradiologists together with the PET-readings (and these data were new information for them) and thus their increased sensitivity. Finally, the quality of MRI images deriving from the new hybrid PET/MRI systems was much better. These factors combined might explain the 60% difference (18 patients/30 nonlesional patients).
Moreover, in nine of the 30 cases, findings of hybrid FDG-PET/MRI resulted in a significant change in decision-making (iEEG monitoring, resective surgery or not eligible for any further invasive procedures) (Figs. 2 and 3).

Lesional group
Hybrid FDG-PET/MRI disclosed at least one new morphological lesion in 9 patients and glucose hypometabolism in 30 patients within the lesional group.
Four patients found to be nonlesional in this study. In 17 cases the original lesion was confirmed. As we mentioned above, the major difference was the application of 3D FLAIR sequence which was lacking in the earlier imaging protocol, which might in part explain the newly identified specific epileptogenic lesions in the lesional group.
Hybrid FDG-PET/MRI has also significantly changed the original therapeutic plans in the lesional group. Three patients became eligible for resective surgerytwo of them were operated with an Engel II outcome. Six patients were considered as not eligible for any further invasive procedure, and 18 cases still remained as candidates for iEEG. iEEG was already performed in 6 of 18 patients concluding in resective surgery with Engel I outcome in 3 cases and Engel II outcome in remaining 3 cases. Two patients remained as candidates for resective surgery -one of them was operated with an Engel I outcome. In both groups the new anatomical or functional lesions were found to be clinically significant.
In the lesional group, hybrid FDG-PET/MRI investigation altered the the epileptologist's original decision in 10 of 30 cases: resective surgery, iEEG, or not eligible for any further invasive procedures (Figs. 4 and 5).

Conclusions
Our study was undertaken to evaluate the potential improvement on decision-making using a hybrid FDG-PET/MRI scanner in epilepsy surgery algorithm, compared to separate 3 T MRI and electroclinical data. The results of hybrid FDG-PET/MRI significantly altered the original plans in 19 of 60 cases. In the nonlesional group, in 18 cases, novel specific epileptogenic MRI-lesions were revealed, proposing potential targets for iEEG monitoring, thus hopefully increasing the chance of successfully identifying the epileptogenic zone in the most difficult epilepsy patients cohort. The patient was nonlesional prior to the study and he/she also remained nonlesional in this study; np: The patient was nonlesional prior to the study and changed to lesional in this study; nc: The patient was suspect for lesional prior to the study but the lesion was not confirmed in this study; pp.: The patient was lesional prior to the study and the study confirmed the original lesion; pp. + : The patient was lesional prior to the study and the study both confirmed the original lesion and found new epileptogenic lesion(s); A.: Positive, revealing unilateral area(s) of hypometabolism in a nonlesional case; B.: Positive, revealing bilateral area(s) of hypometabolism in a nonlesional case; C.: Positive, ipsilateral, related to the MRI-identified lesion in a lesional case; D.: Positive, ipsilateral, but not related to the MRI-identified lesion, pointing to new area(s) within the same hemisphere in a lesional case; E.: Positive, contralateral to MRI-lesion in a lesional case; F.: Positive, bilateral in lesional case in a lesional case; G.: Negative; 1.: Remained as iEEG candidate; 2.: Resective surgery is available instead of iEEG; 3.: Considered as not eligible for any further invasive procedures instead of iEEG; 4.: Became iEEG candidate instead of resective surgery; 5.: Considered as not eligible for any further invasive procedures instead of resective surgery; 6.: Remained as candidate for resective surgery study design, writing the application for ethical permission, manuscript preparation. ZsH: patients analysis. PH: senior epileptologist, patients analysis, semiology. VJ: patients analysis, semiology. CsGy: patients analysis, semiology. BB: patients analysis, semiology. DK: patients analysis, semiology. RH: patients analysis, semiology. FN: patients preparation for PET/MRI examination; and follow-up. AK: patients analysis, semiology. ZsJ: patients analysis, semiology. AU: patients analysis, semiology. KH: manuscript design and correction. JI: manuscript design and correction. EP: manuscript preparation. AF: PET analysis. DF: patient analysis, semiology, study design, manuscript preparation. ZsV: expert of MRI analysis, study design; manuscript preparation. The author(s) read and approved the final manuscript.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval and consent to participate 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. This study was approved by Scientific Research Ethics Committee of the Medical Research Council of Hungary (No. 008899/2016/OTIG). Written informed consent was obtained from all participants.

Consent for publication
Not applicable.