Optimal metrics for SUV values using 18F-FDG PET/MRI in patients with drug-resistant epilepsy and verified epileptogenic lesion

Full Text

Introduction

Epilepsy is one of the most common neurological disorders, affecting approximately 50 million people of all ages worldwide [1]. Despite the availability of numerous antiepileptic agents, 30% of patients are resistant to pharmacological therapy. These patients are at higher risk of seizures, sudden death, injuries, psychosocial dysfunction, and reduced quality of life [2–4]. For patients meeting the criteria for drug-resistant epilepsy established by the International League Against Epilepsy (ILAE), surgery is the most effective treatment option [3]. Surgical treatment of patients with drug-resistant epilepsy and concordant magnetic resonance imaging (MRI) and electroencephalography (EEG) findings is effective in 60–70% of cases, depending on the type of treatment and population [5]. The success of surgical treatment also depends directly on the diagnostic accuracy of localizing the epileptogenic area. Diagnostic challenges primarily arise in patients with diffuse or multifocal brain lesions, as well as in patients with MRI-negative epilepsy and discordant MRI and EEG findings [3, 6–8]. The gold standard for diagnosis is intracranial EEG; however, in complex diagnostic cases, the use of additional minimally invasive techniques can improve the effectiveness of surgical treatment. These include high-resolution MRI and functional MRI, single-photon emission computed tomography (SPECT), positron emission tomography (PET) combined with computed tomography (CT), as well as the novel hybrid method PET/MRI [9, 10].

The hybrid PET/MRI method employing the radiopharmaceutical ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) is multimodal, minimally invasive, and painless; it has high reproducibility, does not require complex or lengthy preparation, has few contraindications and possible adverse reactions, involves low radiation exposure, and in most cases does not require sedation. According to the literature, PET/MRI demonstrates higher sensitivity in diagnosing structural focal epilepsy compared to PET, PET/CT, and MRI [11–13].

Detecting the area of ¹⁸F-FDG hypometabolism in the epileptogenic focus during PET is traditionally based on visual analysis; analysis techniques for hypometabolism in the epileptogenic focus have not been standardized and remain a subject of scientific investigation.

Materials and methods

A total of 10 healthy volunteers and 19 patients with focal drug-resistant epilepsy (15 patients with temporal lobe epilepsy and 4 patients with extratemporal epilepsy) receiving treatment at the Federal Center for Brain Research and Neurotechnologies of the FMBA of Russia in 2024 underwent ¹⁸F-FDG brain PET/MRI. All patients were receiving antiepileptic agents in polytherapy regimens with no effect or only partial effect. All patients underwent clinical neurological examination, prolonged video-EEG monitoring, hybrid ¹⁸F-FDG PET/MRI brain scanning, and analysis of functional, structural, and metabolic changes in the brain.

Prolonged video-EEG monitoring was performed using Neuron-Spectr-4/P and Neuron-Spectr-65 devices (Neurosoft) in accordance with the recommendations of the International Federation of Clinical Neurophysiology, using 21 electrodes placed according to the 10–20 system with additional ECG channels. During video-EEG monitoring, the bioelectrical activity of the brain was studied in states of active and passive wakefulness, with standard functional tests (photic stimulation — rhythmic light stimulation with increasing frequency from 2 to 50 Hz with the light source at 30 cm from the eyes; and hyperventilation for 3 minutes). The study duration ranged from 24 to 48 hours, and sleep recording was performed during video-EEG monitoring.

PET/MRI was performed using an integrated SIGNA PET/MR system (GE Healthcare), which allows simultaneous PET and MR imaging of the brain. For all PET/MRI studies, an 8-channel head coil for high-resolution neuroradiological studies was used. In all cases, patients were observed by medical staff for at least 2 hours prior to the study; if an epileptic seizure occurred, the study was rescheduled to another day. Before ¹⁸F-FDG PET/MRI of the brain, standard preparation for ¹⁸F-FDG PET studies was performed [14]. Patients received intravenous injection of the ¹⁸F-FDG-based radiopharmaceutical at a dosage determined by the radiologist based on patient weight and height (125–250 MBq, mean 180 MBq). After radiopharmaceutical injection, patients remained in a darkened room under medical staff observation for 20–30 minutes, maintained a comfortable supine position with eyes closed, and refrained from using mobile devices, listening to music, reading, or active movements.

During PET data acquisition (over 15 minutes, matrix 192 × 192), standard MRI sequences were obtained based on the technical capabilities of the system and the recommended MRI protocol for epilepsy patients — HARNESS MRI. According to the ILAE 2019 recommendations, this protocol includes a set of isotropic 3D high-resolution pulse sequences: T1, T2, and T2-FLAIR with a minimum voxel size of 1 × 1 × 1 mm³, as well as 2D T2-weighted images in the coronal plane [15]. In addition to the sequences listed above, an attenuation correction sequence lasting 18 seconds was also acquired. This sequence is used to generate atlas-based correction maps for pseudo-CT, which provides information on continuous attenuation for the head using a single-head CT-based atlas.

Two nuclear medicine physicians and two neuroradiologists independently performed visual analysis of the acquired hybrid imaging studies. The images were also analyzed using an Advantage Workstation 4.6 (GE Healthcare) with specialized clinical software — CortexID Suite v. 1.04-5. PET-MRI image analysis was performed sequentially: first, MRI images were evaluated and structural changes in brain parenchyma were analyzed. Subsequently, qualitative and quantitative assessment of brain metabolism was performed based on PET data, evaluating asymmetry of ¹⁸F-FDG metabolism in corresponding regions of the brain. Quantitative analysis included measurement of standardized uptake values (SUV_max, SUV_mean, and SUV_peak) in regions of interest and in attenuation-corrected regions. Thereafter, in conjunction with a neuroradiologist, structural changes identified on MRI were compared with metabolic areas identified on PET. In cases of discordance between structural brain changes and areas of hypometabolism, a repeat detailed analysis of structural brain changes in the identified ¹⁸F-FDG hypometabolism zones was performed.

All patients were discussed at a multidisciplinary conference attended by epileptologists, neurophysiologists, nuclear medicine physicians, neuroradiologists, and neurosurgeons. At the conference, clinical data were compared with the results of the performed studies: prolonged video-EEG monitoring and PET/MRI. Based on the conference findings, a decision was made regarding further patient management: surgical resection of the epileptogenic focus (6 patients), invasive video-EEG monitoring (13 patients), of whom 4 also underwent surgical resection of the epileptogenic focus, while 9 patients with bilateral localization of epileptogenic foci did not undergo resection but received other types of surgical treatment — vagus nerve stimulation or deep brain stimulation. In all cases where surgical resection was performed, a pathological examination was conducted.

Results

All cases patients were conditionally divided into two groups. The first group comprised 10 patients in whom the localization of the epileptogenic area was confirmed by surgical resection of the epileptogenic focus, with Engel class I–II outcomes at 12 or more months of postoperative follow-up (follow-up period ranged from 12 to 25 months, mean 15.7 months). This group consisted of diagnostically complex patients. In two patients, no MRI abnormalities were detected, and the seizure onset zone was not identified on prolonged video-EEG monitoring. However, on ¹⁸F-FDG PET/MRI, an area of ¹⁸F-FDG hypometabolism was identified in the temporal lobe. Pathological examination revealed mesial temporal sclerosis (MTS) in one case and focal cortical dysplasia (FCD) type 1b in the other. In three patients, MRI identified two or more potential epileptogenic foci: MTS and meningocele in the temporal lobes; MTS and multinodular and vacuolating neuronal tumor (MVNT) in the temporal lobes; and multiple subependymal heterotopia nodules in the temporal and occipital lobes of one hemisphere. On prolonged video-EEG monitoring, in the first case, potential epileptogenic foci were identified in both temporal lobes; in the other two cases, foci were identified in one of the temporal lobes. On ¹⁸F-FDG PET/MRI, an area of ¹⁸F-FDG hypometabolism was identified in one temporal lobe in two patients, and the structural changes identified on MRI were confirmed by pathological examination. In the remaining five patients, MRI identified one potential epileptogenic focus (unilateral MTS in four patients, FCD type II of the frontal lobe in one patient). In all five cases, prolonged video-EEG monitoring did not identify a seizure onset zone, or epileptic activity was observed in both hemispheres. On ¹⁸F-FDG PET/MRI, a single area of ¹⁸F-FDG hypometabolism was identified in all five patients, and its localization corresponded to the MRI findings; the structural changes identified on MRI were confirmed by pathological examination (Figures 1, 2).

 

Fig. 1. ¹⁸F-FDG PET/MRI of the brain in a patient with FCD type IIb.

A — сoregistered PET/MRI image in the coronal plane showing an area of ¹⁸F-FDG hypometabolism in the medial aspects of the right frontal lobe (circle); B — T2-weighted image in the coronal plane showing cortical thickening and reduced gray-white matter demarcation in the medial aspects of the right frontal lobe, within the area of ¹⁸F-FDG hypometabolism (circle); C — сoregistered PET/MRI image in the axial plane showing an area of ¹⁸F-FDG hypometabolism in the medial aspects of the right frontal lobe (circle); D — T2-FLAIR image in the axial plane showing cortical thickening and reduced gray-white matter demarcation in the medial aspects of the right frontal lobe, within the area of ¹⁸F-FDG hypometabolism (circle).

 

Fig. 2. ¹⁸F-FDG PET/MRI of the brain in a patient with MTS.

A — сoregistered PET/MRI image in the axial plane showing an area of ¹⁸F-FDG hypometabolism in the mediobasal aspects of the right temporal lobe (arrow); B — T2-FLAIR image in the axial plane showing increased MRI signal intensity from the right hippocampus and reduced volume of the right hippocampus (arrow); C — ¹⁸F-FDG PET image in the axial plane showing an area of ¹⁸F-FDG hypometabolism in the mediobasal aspects of the right temporal lobe (arrow); D — coregistered PET/MRI image in the coronal plane showing an area of ¹⁸F-FDG hypometabolism in the mediobasal aspects of the right temporal lobe (arrow); E — T2-FLAIR image in the coronal plane showing increased MRI signal intensity from the right hippocampus and reduced volume of the right hippocampus (arrow); F — T2-weighted image in the coronal plane showing increased MRI signal intensity from the right hippocampus and reduced volume of the right hippocampus (arrow).

 

Thus, in the group of operated patients, the use of ¹⁸F-FDG PET/MRI allowed identification of the epileptogenic area localization in two MRI-negative patients, clarified the localization of the epileptogenic area and ruled out other potential epileptogenic foci in three patients with bilateral and multifocal involvement, and confirmed MRI findings in 5 patients with discordant or indeterminate EEG data.

The least pronounced reduction in semi-quantitative ¹⁸F-FDG metabolism parameters was recorded in patients with FCD (n = 2): SUV_max = 8.7 ± 0.5 (range: 8.1–9.2), SUV_mean = 5.4 ± 0.3 (range: 5.1–5.6), SUV_peak = 5.4 ± 0.2 (range: 5.2–5.5). However, the small sample size does not permit generalizable conclusions.

In patients with MTS (n = 6), a greater variability in semi-quantitative ¹⁸F-FDG metabolism parameters was observed: SUV_max = 6.1 ± 1.1 (range: 4.2–7.6), SUV_mean = 4.3 ± 0.8 (range: 3.2–5.8), SUV_peak = 4.7 ± 0.5 (range: 3.8–5.3). The mean values of all parameters were lower than those in patients with FCD, which may reflect more pronounced reduction in metabolic activity in MTS foci. The moderate data variability indicates heterogeneity of metabolic characteristics within this group.

In the patient with MVNT (n = 1), lower values of all analyzed parameters were recorded compared to the mean values in the other groups: SUV_max = 5.2, SUV_mean = 3.2, SUV_peak = 3.6. However, as these data are from only one patient, no conclusions can be drawn.

The second group comprised 9 patients in whom the localization of the epileptogenic area was confirmed by invasive video-EEG monitoring. In all cases, invasive video-EEG monitoring was performed after PET/MRI, and surgical resection of the epileptogenic focus was not performed. The group consisted of diagnostically challenging patients. In 3 patients, no structural changes were detected on MRI, and the seizure onset area was not identified on scalp video-EEG monitoring. On PET/MRI, an area of ¹⁸F-FDG hypometabolism was identified in one of the temporal lobes. In two of these patients, upon repeat targeted analysis of MRI data within the ¹⁸F-FDG hypometabolism area, a slight increase in MRI signal intensity on the FLAIR sequence was noted — originating from the hippocampal body in one case and from the amygdala in the other — without changes in the volume of these structures; these findings were interpreted as manifestations of MTS. In 3 patients, MRI revealed features of MTS: in one patient, the seizure onset area was not identified on scalp video-EEG monitoring; in the other two cases, MRI and EEG findings were discordant (unilateral involvement on MRI with bilateral temporal epileptic activity on EEG in one case; bilateral involvement on MRI with unilateral temporal epileptic activity on EEG in the other). In 2 patients, similar structural changes were identified in the temporal lobes (cavernous malformations and meningoencephaloceles); the seizure onset area was not identified on scalp video-EEG monitoring. In one patient, postoperative changes were noted in the medial aspects of one frontal lobe following resection of FCD; the seizure onset area was not identified on scalp video-EEG monitoring.

According to invasive video-EEG monitoring, in 8 of the 9 patients, seizure onset areas were independently recorded in the mediobasal aspects of the temporal lobes, and the localization of the area with the greatest representation of epileptic seizures corresponded to the localization of the ¹⁸F-FDG hypometabolism zone identified on PET/MRI. Of these, in one MRI-negative patient, stereo-EEG also recorded epileptic seizures with a diffuse onset. In one patient with prior resection of FCD in the medial aspects of the frontal lobe, invasive video-EEG monitoring recorded seizure onset areas independently in both cingulate gyri: at the periphery of the postoperative cavity and contralaterally. On PET/MRI, against ¹⁸F-FDG hypometabolism in the area of postoperative changes in one cingulate gyrus, it was not possible to reliably determine the localization of residual FCD tissue; no structural or metabolic changes were identified in the contralateral hemisphere.

Based on the multidisciplinary conference findings, vagus nerve stimulator implantation was recommended for 8 patients. In one case (the patient with diffuse onset of epileptic seizures), a destructive procedure on the epileptic foci was recommended, with prior brain mapping based on invasive implantation of epidural electrodes and monitoring using focused ultrasound.

During the study, a statistical analysis of ¹⁸F-FDG PET/MRI of the brain was performed on two groups of patients with confirmed epileptogenic areas. In all cases, the parameters SUV_max, SUV_mean, and SUV_peak were calculated for the epileptogenic focus, both as absolute values and as the percentage ratio of the hypometabolic area to contralateral brain regions.

Normality was assessed using the Shapiro–Wilk test for each parameter. For Group 1: SUV_max (W = 0.962; p > 0.05), SUV_mean (W = 0.947; p > 0.05), and SUV_peak (W = 0.953; p > 0.05). Despite slight skewness and kurtosis, these data do not provide grounds to reject the hypothesis of a normal distribution. For Group 2: SUV_max (W = 0.877; p > 0.05) — the Shapiro–Wilk test formally does not reject the hypothesis of a normal distribution, but high skewness (~1.12) and elevated kurtosis (~3.9) indicate deviation from normality due to an outlier (SUV_max = 10.6). For SUV_mean in Group 2 (W = 0.852; p > 0.05), the Shapiro–Wilk test also does not reject the hypothesis of a normal distribution; skewness (~1.35) and kurtosis (~6.2) indicate deviation due to an outlier (SUV_mean = 6.3). For SUV_peak in Group 2 (W = 0.858; p > 0.05), the hypothesis of a normal distribution is likewise not rejected, but high skewness (~1.05) and elevated kurtosis (~3.7) suggest a possible outlier (SUV_peak = 6.9). For Group 3: SUV_max (W = 0.943; p > 0.05), SUV_mean (W = 0.951; p > 0.05), and SUV_peak (W = 0.946; p > 0.05).

Based on the normality of the parameter distributions for each group, mean values and standard deviations were calculated for each parameter (Table 1), enabling comparison between groups regarding the degree of hypometabolism and assessment of differences between them.

 

Table 1. Mean values and standard deviations of percentage ratios of SUV_max, SUV_mean, and SUV_peak between the hypometabolic area and contralateral regions in patients

Parameter	SUVmax	SUVmean	SUVpeak
Group 1 (n = 10)
mean value, %	19.4	18.4	19.7
standard deviation	12.7	9.8	11.5
Group 2 (n = 9)
mean value, %	14.1	12.9	11.2
standard deviation	7.6	5.0	7.5
Group 3 (n = 10)
mean value, %	5.6	6.1	7.6
standard deviation	3.5	5.1	6.0

 

For each of the three relative metrics (%SUV_max, %SUV_mean, and %SUV_peak), calculated as the ratio of SUV in the focus to contralateral SUV, a comparison was performed across the three groups.

For %SUV_max, the medians and interquartile ranges were as follows: in Group 1 — 14.5% (10.3–33.3%), in Group 2 — 12% (8–18%), and in Group 3 — 5.5% (2.5–8.5%). The difference between Group 1 and Group 3 was statistically significant (p = 0.006) with a large effect size (Cohen’s d ≈ 1.40); the same was true for Group 2 compared to controls (p = 0.008; d ≈ 1.36). Despite the wider spread in Group 1, the absolute shift in medians shows a clear gradient from pronounced metabolic reduction in patients to the lowest values in controls.

The %SUV_mean parameter demonstrated the least degree of distribution overlap: medians — 14.5% (10–25.3%) in Group 1, 11% (10–12%) in Group 2, and only 3.5% (2.3–9.5%) in Group 3. Here, the differences were most significant: p = 0.003 (d ≈ 1.51) when comparing Group 1 with Group 3, and p = 0.024 (d ≈ 1.29) for Group 2. The narrow IQRs with such large median differences from controls may indicate high reproducibility and sensitivity specifically for this parameter.

For %SUV_peak, we observed intermediate values: medians of 20% (9.3–24.5%) in the operated group, 9% (5–14%) in the invasive monitoring group, and 5% (2.5–11.0%) in the control group. The differences were significant, but the effect size was somewhat lower: d ≈ 1.25 (p = 0.012) for Group 1 vs. Group 3 and d ≈ 1.12 (p = 0.018) for Group 2 vs. Group 3.

In Group 1, the highest mean values were observed across all three parameters, which may indicate a pronounced hypometabolic focus corresponding to the epileptogenic area, subsequently confirmed by pathological examination and reduced seizure frequency in the postoperative period. In Group 2, a certain reduction in mean percentage ratios was observed across all parameters compared to Group 1, likely due to the presence of a contralateral epileptogenic focus detected by intracranial EEG.

Analysis of brain PET/MRI data from patients with MRI-negative (n = 5) and MRI-positive (n = 14) epilepsy revealed differences in the metabolic activity of foci. In the MRI-negative group, mean values were: SUV_max = 6.3 ± 1.2 (range: 4.7–8.1), SUV_mean = 4.0 ± 0.8 (2.9–5.1), SUV_peak = 4.5 ± 0.8 (3.5–5.5). In the MRI-positive group, values were higher and more variable: SUV_max = 6.9 ± 2.0 (range: 4.3–10.6), SUV_mean = 4.2 ± 1.2 (2.7–6.3), SUV_peak = 4.8 ± 1.1 (3.1–6.9). Despite the absence of statistically significant differences (p > 0.05 for all parameters, Mann–Whitney U test), in the MRI-positive group, one patient with MTS showed extreme values (SUV_max = 10.6), which may indicate individual characteristics of ¹⁸F-FDG metabolism.

SUV_max in the MRI-positive group demonstrates greater variability (± 2.0 vs. ± 1.2). SUV_mean values are comparable between the two groups (4.0 vs. 4.2), but in MRI-negative cases, the minimum values are higher (2.9 vs. 2.7), although this does not reach statistical significance due to the small sample size. In the MRI-positive group, a strong positive correlation was found between SUV_max and SUV_peak (r = 0.89; p < 0.001), confirming measurement consistency. In the MRI-negative group, the SUV_max–SUV_peak correlation was weaker (r = 0.72; p = 0.07), likely due to the limited number of observations.

Analysis of PET/MRI data from patients with temporal lobe epilepsy (n = 17) and extratemporal epilepsy (n = 2) also revealed differences in metabolic parameters. In the temporal lobe epilepsy group, mean values were: SUV_max = 6.6 ± 2.0 (range: 4.3–10.6), SUV_mean = 3.9 ± 1.0 (2.7–6.3), SUV_peak = 4.7 ± 1.1 (3.1–6.9). In the extratemporal epilepsy group, higher mean values were recorded: SUV_max = 7.4 ± 1.6 (range: 6.0–9.2), SUV_mean = 4.6 ± 1.0 (3.7–5.6), SUV_peak = 4.7 ± 0.7 (3.9–5.2); however, the small sample size (n = 2) does not permit statistically significant conclusions.

SUV_max in patients with temporal lobe epilepsy demonstrates wide variability (± 2.0), including an extreme value of SUV_max = 10.6 in one patient, whereas in the extratemporal group, the maximum SUV_max value was 9.2 (in a patient with FCD type IIb) with less variability. SUV_mean values were higher in the extratemporal group (4.6 ± 1.0 vs. 3.9 ± 1.0), which may reflect the normally lower ¹⁸F-FDG metabolism in the temporal lobes. In the temporal lobe epilepsy group, a strong positive correlation was found between SUV_max and SUV_peak (r = 0.91; p < 0.001), similar to previous analyses. In the extratemporal group, the SUV_max–SUV_peak correlation was also high (r = 0.98), but statistical significance was not calculated due to the small sample size; therefore, the analysis for the extratemporal group is relative.

Discussion

The radiopharmaceutical ¹⁸F-FDG, widely used for PET diagnostics in epilepsy, is readily available, has a half-life of 110 minutes, and accumulates in tissues proportional to glucose metabolism. The mechanism underlying the formation of the ¹⁸F-FDG hypometabolism area in the epileptogenic focus remains poorly understood and may be multifactorial. One theory explains the reduction in glycolytic activity as a consequence of inhibitory mechanisms of synaptic plasticity; the same mechanism may also account for hypometabolism in the ipsilateral frontal and parietal lobes and the basal ganglia. Identified contralateral areas of ¹⁸F-FDG hypometabolism may represent additional pathological foci, including epileptogenic foci, and require correlation with MRI and EEG data. PET scanning is performed during the interictal period. This is because adequate concentration of ¹⁸F-FDG in brain parenchyma is achieved 30–40 minutes after injection and reflects cumulative metabolism over this period. Such a long radiopharmaceutical uptake time precludes assessment of brief neuronal events, such as an epileptic seizure, which in most cases lasts one or several minutes [16]. The methodology for performing ¹⁸F-FDG PET is generally standardized: patients must remain in a relaxation room (quiet, darkened space) for at least 15 minutes before radiopharmaceutical administration and for 30 minutes thereafter. The European Association of Nuclear Medicine guidelines note the necessity of performing video-EEG monitoring for at least 20 minutes before and at least 20 minutes after radiopharmaceutical administration in order to avoid imaging of ¹⁸F-FDG hypermetabolism areas due to the ictal phase, as well as postictal areas of ¹⁸F-FDG hypometabolism that are not directly related to the epileptogenic focus. During the interictal period, the epileptogenic focus is identified on PET as a hypometabolism area; however, reduced ¹⁸F-FDG accumulation in potential foci is nonspecific for epilepsy. Therefore, correlating PET data with MRI or CT imaging to visualize structural changes within ¹⁸F-FDG hypometabolism areas is crucial [17].

In most comparative analyses, the size of the focus detected by PET exceeds that of the area of altered MRI signal identified on MRI [18]. The ¹⁸F-FDG hypometabolism area likely includes both the area of seizure initiation and the area of propagation; therefore, PET allows lateralization and approximate localization of the epileptogenic area, but precise determination of the seizure onset region using PET alone — particularly for small foci — may be challenging [19]. In some cases, topographic localization of the hypometabolism area by PET can help re-evaluate MRI data in patients with MRI-negative epilepsy and identify structural changes within the ¹⁸F-FDG hypometabolism area, which is particularly relevant for small focal cortical dysplasias and hippocampal sclerosis not accompanied by significant hippocampal volume loss [20]. Some detected ¹⁸F-FDG hypometabolism foci are not confirmed by MRI; nevertheless, when the localization of the epileptogenic area is confirmed by intracranial EEG, surgical resection of this region leads to clinical improvement in most cases, up to complete seizure freedom. Analysis of PET data together with MRI, when these modalities are concordant, can identify up to 89% of patients with favorable surgical outcomes [21]. Discordance between the PET hypometabolism area localization, the epileptogenic area identified by EEG, and the clinical semiology may conversely predict unfavorable surgical outcomes [22]. Detection of the ¹⁸F-FDG hypometabolism area in the epileptogenic focus is traditionally based on visual analysis; quantitative analysis techniques for hypometabolism in the epileptogenic focus have not been standardized and remain a subject of scientific investigation.

Analysis of brain PET/MRI data in the group of operated patients revealed features of metabolic activity depending on the histological diagnosis. In patients with MTS (n = 6), the smallest reduction in semi-quantitative ¹⁸F-FDG metabolism parameters was recorded, which may be attributable to chronic inflammation with moderate parameter variability. In the two patients with FCD included in the study, the greatest variability in semi-quantitative SUV_max and SUV_mean parameters was noted; however, the small patient sample does not permit statistically reliable conclusions from this observation, and verification of the results requires expansion of the study cohort for the histologically rare entities. MTS occupies an intermediate position with moderate data variability, which may be related to differences in disease duration and phase among the included patients, as well as variability in structural hippocampal changes (from MRI-negative to pronounced volume loss). In patients with FCD and MTS, the smallest reduction in SUV_max values was observed; for these patient groups, the SUV_mean parameter was the best for localizing the epileptogenic focus. The lowest values of all analyzed parameters were observed in the patient with multinodular and vacuolating neuronal tumor. Given the small sample size, verification of the results requires expansion of the study cohort.

Analysis of brain PET/MRI data from patients with MRI-negative and MRI-positive epilepsy revealed no statistically significant differences in semi-quantitative ¹⁸F-FDG metabolism parameters. Thus, the use of PET/MRI both confirms structural changes detected on MRI that may serve as a potential epileptogenic substrate and enables identification of the epileptogenic focus in patients without structural brain pathology. Upon targeted re-evaluation of MRI images within the area of ¹⁸F-FDG PET hypometabolism, small structural changes of the epileptogenic focus can be detected. On the one hand, this approach allows identification of small structural changes in the cerebral cortex and hippocampi that might be missed on initial analysis. On the other hand, detecting a venous developmental anomaly, cystic-gliotic changes, or other structural alterations in the hypometabolic area — changes that are traditionally associated with ¹⁸F-FDG hypometabolism but may not themselves represent the epileptogenic substrate — permits а more cautious interpretation of metabolic changes in brain parenchyma. Of note, determining the localization of the epileptogenic focus may be challenging in the setting of an extensive ¹⁸F-FDG hypometabolism area due to postoperative or post-inflammatory changes.

In the group of patients with temporal lobe epilepsy, lower SUV_max and SUV_mean values were recorded, which may reflect the normally lower ¹⁸F-FDG metabolism in the temporal lobes compared to the cortex of other brain regions [23]. Because visual and semi-quantitative analysis of ¹⁸F-FDG metabolism is based on comparison with contralateral symmetric brain regions, detection of less pronounced reduction in ¹⁸F-FDG metabolism in a contralateral epileptogenic focus may be challenging. In patients with temporal lobe epilepsy and bilateral localization of epileptogenic foci, visual and semi-quantitative analysis of ¹⁸F-FDG PET/MRI images allowed identification of the epileptogenic area with the greatest representation of epileptic seizures but did not permit complete exclusion of a contralateral focus.

Conclusion

PET/MRI is a novel hybrid diagnostic modality that can be incorporated into the presurgical evaluation algorithm for patients with drug-resistant epilepsy and has the potential to improve the detection of the epileptogenic focus in complex diagnostic cases, particularly in patients with MRI-negative epilepsy and in those with discordant MRI and video-EEG findings.

Statistical analysis of quantitative ¹⁸F-FDG metabolism parameters in the epileptogenic focus confirmed by invasive methods suggests that the SUV_mean parameter is the optimal metric for identifying hypometabolic areas suspicious for the epileptogenic focus, due to more homogeneous quantitative values within the sample and lower standard deviations, particularly in patients with FCD and MTS.

About the authors

Tatiana M. Rostovtseva

Federal Center of Brain Research and Neurotechnologies

Author for correspondence.
Email: rostovtsevat@mail.ru
ORCID iD: 0000-0001-6541-179X

radiologist, Department of X-ray and radionuclide diagnostic methods

Russian Federation, Moscow

Mikhail B. Dolgushin

Federal Center of Brain Research and Neurotechnologies

Email: dolgushin.m@fccps.ru
ORCID iD: 0000-0003-3930-5998

Dr. Sci. (Med.), Prof. of RAS, Head, Department of X-ray and radionuclide diagnostic methods

Russian Federation, Moscow

Rostislav V. Nadelyaev

Federal Center of Brain Research and Neurotechnologies

Email: nadelaevr@gmail.com
ORCID iD: 0009-0005-7367-9311

radiologist, Department of X-ray and radionuclide diagnostic methods

Russian Federation, Moscow

Andrey V. Dvoryanchikov

Federal Center of Brain Research and Neurotechnologies

Email: dvoryanchikov.a@fccps.ru
ORCID iD: 0009-0009-0678-7821

medical physicist, Department of X-ray and radionuclide diagnostic methods

Russian Federation, Moscow

Yulia V. Rubleva

Federal Center of Brain Research and Neurotechnologies

Email: rubleva@fccps.ru
ORCID iD: 0000-0002-3746-1797

Cand. Sci. (Med.), Head, Neurological department No. 1

Russian Federation, Moscow

Vidzhai M. Dzhafarov

Federal Center of Brain Research and Neurotechnologies

Email: djafarov.v@fccps.ru
ORCID iD: 0000-0002-5337-8715

Cand. Sci. (Med.), neurosurgeon, Neurosurgical department

Russian Federation, Moscow

Ilya V. Senko

Federal Center of Brain Research and Neurotechnologies

Email: senko@fccps.ru
ORCID iD: 0000-0002-5743-8279

Dr. Sci. (Med.), Head, Neurosurgical department

Russian Federation, Moscow

Elena A. Baranova

Federal Center of Brain Research and Neurotechnologies

Email: baranova.e@fccps.ru
ORCID iD: 0000-0002-9200-9234

Cand. Sci. (Med.), Head, Department of functional diagnostics of nervous system diseases

Russian Federation, Moscow

Yulia A. Voronkova

Federal Center of Brain Research and Neurotechnologies

Email: voronkova.y@fccps.ru
ORCID iD: 0000-0003-3682-5736

neurophysiologist, Department of functional diagnostics of nervous system diseases

Russian Federation, Moscow

Olga I. Patsap

Federal Center of Brain Research and Neurotechnologies

Email: patsap.o@fccps.ru
ORCID iD: 0000-0003-4620-3922

Cand. Sci. (Med.), Head, Pathomorphological department

Russian Federation, Moscow

Sergey G. Burd

Federal Center of Brain Research and Neurotechnologies

Email: burd.s@fccps.ru
ORCID iD: 0000-0001-6256-2576

Dr. Sci. (Med.), Prof., Head, Epilepsy and paroxysmal disorders department

Russian Federation, Moscow

References

Supplementary files