Patients were enrolled following the inclusion criteria: subjects who underwent preoperative evaluations with SEEG, subjects who finished resective surgeries from January 2016 to December 2018 in our epilepsy centers in Beijing, patients who had met the criteria of medication-resistant epilepsy for no less than 2 years, and subjects who had previously been diagnosed with TSC in accordance with the revised diagnostic criteria of Northrup (Northrup and Krueger, 2013). The exclusion criteria included subjects with one to three cortical tubers; subjects with obvious lymphangioleiomyomatosis, renal angiomyolipomas, and cardiac rhabdomyomas; and patients with serious cardiac, renal, or lung dysfunction (Liang et al., 2010, 2017). This study was approved by the Ethics Committee of the Fourth Medical Center, General Hospital of PLA, and the written consent was not signed for a retrospective study.

Non-invasive preoperative evaluations included neurological history (e.g., clinical seizure semiology) and physical examination, MRI, long-term scalp video EEG recordings, PET, and neuropsychological testing. MRI scans included 3.0-T routine axial T1-weighted, T2-weighted, and diffusion-weighted imaging; sagittal T1-weighted imaging; and 1-mm thickness by zero interval axial and coronal T2–fluid-attenuated inversion recovery (FLAIR) imaging. The number of cortical tubers was counted using axial T2-FLAIR images. Neuropsychological tests included Wechsler Intelligence Scale IV (Chinese revision) for measuring intelligence quotient and the overall subscale of quality of life on the Quality of Life in Childhood Epilepsy Questionnaire.

Stereo-electroencephalograph electrodes with 8–16 contacts, 0.8 mm in diameter, 2 mm in length for contacts, and 1.5 mm in intercontact interval (Huake Company, Beijing, China) were embedded under generalized anesthesia for recording intracranial EEGs, in order to localize the epileptogenic tubers in TSC patients with multiple potential epileptogenic cortical tubers. The SEEG electrodes covered the potential epileptogenic cortical tubers, which had calcifications or cystic changes on MRI, had abnormal findings on PET images, or were localized in regions with focal ictal symptoms or focal ictal and/or interictal epileptiform discharges on scalp EEGs (Liang et al., 2017). In addition, the adjacent cortexes to those tubers were also covered with SEEG electrodes (Yu et al., 2019).

Data from a minimum of five habitual seizures episodes were required for further analysis and identification. The epileptogenic tuber was identified as the first tuber with initial rhythmical discharge on SEEG before clinical seizure attack. Propagating tubers were identified by secondary rhythmical discharges on SEEG before or after clinical seizure attack in 10 s after ictal EEG onset. If more than one tuber exhibited an initial rhythmical discharge on SEEG during the same seizure or during different seizure episodes, comprehensive analysis was performed by combining MRI-PET image fusion data and clinical semiology to distinguish independent epileptogenic tubers from propagating tubers (Yu et al., 2019; Liu et al., 2020).

Interictal SEEG signals were recorded with an EEG acquisition system (Natus, United States) with a sampling frequency of 0–4,000 Hz (Ren et al., 2018). Five segments of 5-min interictal SEEGs during slow-wave sleep at midnight were used to analyze the occurrence frequency of FRs. Those segments were separated from each other and from seizure episodes by a minimum of 2 h. Slow-wave sleep was defined by the presence of more than 25% delta activity in 30-s epochs by visual inspection (Urrestarazu et al., 2007).

Bipolar montage with pairs of two adjacent EEG electrodes successively connected was used. The reference electrodes were excluded from the dataset. The automated detection of FRs was performed using software and previously described methods (Ren et al., 2018). During analysis, each contact and tuber with FRs were counted for every epoch and then averaged for a 5-min interval. The occurrence frequency of FRs was described with the number of FRs and the rate of FRs in each contact. The FR rate was calculated using the following formula: (the number of FRs in this contact/the maximum number of FRs among all contacts of this patient). A contact was defined as an FR contact when the occurrence frequency of FRs was more than 0.2/min in this contact. Similarly, a tuber with no less than one FR contact was defined as an FR tuber. In addition, when the FR rate was more than 0.5 in a certain contact (Burnos et al., 2014), we defined it as a high FR rate contact. Also, if no less than one high FR rate contact occurred in a tuber, we defined the tuber as a high FR rate tuber.

According to the positional relation of contacts and tubers, the locations of contacts were defined as inner of the tuber, junction region of the tuber, and out of the tuber. The inner of the tuber was defined as inner three-fourths of the tuber. The junction region of the tuber was defined as the outer quarter area of the tuber border on MRI-T2-FLAIR plus the adjacent cortical area of gyrus with tuber involvement. Out of the tuber means the cortex of gyrus without tuber involvement (Figure 1).

Patients underwent either lobectomies or tuber resections. Tuber resections were used for epileptogenic tubers within or close to eloquent areas. Lobectomies were applied when epileptogenic tubers were in the anterior temporal lobe or the frontal pole. Multiple tuber resections or lobectomies with tuber resection were considered in cases of multiple epileptogenic tubers unable to be removed by a single lobectomy. Pharmaceutical treatments were provided to all postoperative patients, utilizing optimized combinations of two to four kinds of antiseizure medications. Potential medications used in postoperative patient care were topiramate, vigabatrin, valproate, levetiracetam, lamotrigine, and oxcarbazepine.

Statistical analysis was completed using the SPSS statistical program (version 19.0; SPSS, Inc., Chicago, IL, United States). Postoperative seizure controls were classified according to the Engel method into class I (seizure free), class II (rare seizures), class III (>90% reduction in seizure frequency), and class IV (<90 reduction in seizure frequency). Outcomes were described with percentages, means, and SD. Univariate analysis of categorical variables was performed using χ² and Fisher exact tests. t-tests and F-tests were used for comparison of continuous variables. When the two-tailed error probability P was less than 0.05, the outcome was considered to be significant.

A total of 46 patients with TSC underwent epilepsy surgery at our hospitals from January 2016 to December 2018. Twenty-one patients received implanted SEEG electrodes and underwent resective surgery, and 17 of them with comprehensive FRs data were included in this study (Table 1). Both female (n = 6) and male (n = 11) patients were presented. Patient ages ranged from 2.9 to 12.6 (mean = 6.13 ± 3.06) years. Types of clinical seizures at onset included generalized epileptic spasms (n = 9), generalized tonic–clonic seizure (n = 3), focal seizure (n = 4), and generalized clonic seizure (n = 1). Seizure frequencies included either daily seizures (n = 14) or weekly seizures (n = 3). Age at seizure onset ranged from 0.2 to 5.9 (mean = 1.34 ± 1.60) years. The durations of preoperative seizures ranged from 2.3 to 9.6 (mean = 4.79 ± 2.37) years. Through observation and counting, each patient had 10.18 ± 3.21 (range = 4–16) cortical tubers.

A total of 25 epileptogenic tubers were identified across all patients. Patients were observed to have a single epileptogenic tuber (n = 9) or two epileptogenic tubers (n = 8); no patient was observed to have more than two epileptogenic tubers. Furthermore, 35 early propagating tubers were identified. Surgical interventions varied case by case and included epileptogenic tuber resection (n = 6), lobectomy or multilobar resection (n = 7), and a combination of lobectomy and tuber resection (n = 4). In total, 52 tubers, including 22 epileptogenic tubers and 30 propagating tubers, were removed. At 1-year follow-up, 12 patients (70.6%) achieved seizure freedom (Engel I), one reached Engel II, and the other four cases reached Engel III–IV seizure control. Significant difference was found in the percentage of total removal of epileptogenic tubers between patients with postoperative seizure freedom and those with postoperative continuous seizure at 1-year follow-up (P = 0.0239) (Table 2).

There were 471 (24.2%) FR contacts detected from the 1,944 contacts of 144 implanted SEEG electrodes, including 75 (3.9%) high FR rate contacts. Moreover, 102 (67.1%) FR tubers were observed across all covered tubers (n = 152) in all patients (Table 1).

The occurrence frequency of FR discharge significantly varied across the tuber anatomy (P < 0.01). There were 62.7% (47 of 75) high FR rate contacts located at the junction region of the tubers, and those contacts without FRs mainly located out of the tubers (n = 703, 47.7% of 1,473). High FR rate contacts included two (0.2% of 834) contacts out of the tubers, 47 (8.4% of 561) contacts in the junction region of the tubers, and 26 (4.7% of 549) contacts in the inner of the tubers (Figure 1). There were significant differences in the percentage of the high FR rate contacts in the three parts of cortical tubers (P = 0.0000).

The tubers and contacts with FRs were compared between patients with Engel I seizure controls and those with Engel II–IV seizure controls. There was no significant difference found in percentage (or number) of FR contacts (or tubers) between those two groups (P > 0.05) (Table 2).

The removed brain tissues included 272 FR contacts (Figure 2) and 48 FR tubers (Figure 3) in all patients. Each patient removed 3–30 FR contacts, and the maximum rates of FRs occurrence frequency in remained contacts ranged from 0 to 1 in different cases (Figure 2). Significant differences were found in seizure freedom between patients who totally removed the high FR rate contacts and those who partially removed the high FR rate contacts (100 vs. 16.7%, P < 0.05) (Table 2).

Each patient removed two to four FR tubers, and the maximum FR rate in the remained tubers ranged from none to one in different cases (Figure 3). Significant differences were found in seizure freedom between those who totally removed high FR rate tubers and cases who partially removed high FR rate tubers (100 vs. 16.7%, P < 0.05) (Table 2).

One patient presented asymptomatic epidural hematoma around an electrode. No complication was identified during FR recording.