Distinct gray matter and metabolic characteristics in hypothalamic hamartoma network with different semiology

Abstract

Objective

Hypothalamic hamartomas (HHs) are developmental malformations associated with focal epilepsy. We investigated the patterns of gray matter morphology and cerebral metabolism in individuals with HHs, with and without focal to bilateral tonic–clonic seizures (FBTCSs), aiming to clarify the accompanying network abnormalities.

Methods

We analyzed magnetic resonance imaging (MRI) and fluorodeoxyglucose positron emission tomography (PET) data from 59 patients with HHs (28 with FBTCSs, 31 without), as well as MRI data from 30 healthy controls (HCs) and PET data from 45 HCs. We assessed gray matter voxel‐based morphometry and quantitative analysis of cerebral glucose uptake in HH patients and controls, with age, sex, and total intracranial volume as covariates, and drew correlations with duration of epilepsy and seizure semiology and frequency.

Results

Compared to HCs, HH patients had significantly increased gray matter volume (GMV) in the ipsilateral amygdala, piriform cortex, hypothalamus, and bilateral temporal cortices; patients with FBTCSs primarily showed increased GMV in the HH stalk, whereas those without FBTCSs showed increased GMV prominently in the amygdala. GMVs of amygdala and piriform cortex were greater and the ipsilateral midtemporal cortex was more hypometabolic the longer the duration of epilepsy and the greater the seizure frequency. No significant GMV or cerebral glucose uptake differences were found between HH patients with and without FBTCSs.

Significance

HH‐related epilepsy is a network disorder characterized by widespread abnormalities beyond the lesion. This highlights the importance of considering the whole network when formulating diagnosis and treatment plans.

Key points.

• Patients with HHs exhibited increased GMV in the ipsilateral amygdala, piriform cortex, hypothalamus, and bilateral temporal cortices, suggesting the involvement of a broader epileptogenic network.

• Patients with FBTCSs showed increased GMV mainly in the hypothalamic attachment area, whereas those without FBTCSs had increased GMV primarily in the amygdala. These semiology‐specific patterns support differential network involvement.

• Longer epilepsy duration and higher seizure frequency were associated with larger GMV in the amygdala and piriform cortex, especially in patients without FBTCSs. These areas may act as key hubs in the epileptogenic network of gelastic and focal impaired awareness seizures.

• FDG‐PET analysis revealed significant hypometabolism in the ipsilateral middle temporal gyrus in HH patients with FBTCSs, potentially reflecting a metabolic substrate of seizure generalization.

• The findings underscore the importance of evaluating extralesional network abnormalities (e.g., amygdala, piriform cortex) using targeted SEEG and z‐score mapping. Early, individualized interventions may help prevent progression of network remodeling and improve surgical outcomes.

1. INTRODUCTION

Hypothalamic hamartomas (HHs) are a rare developmental malformation commonly associated with intractable epilepsy and precocious puberty. ¹ Patients with HHs may have range of seizure types, with gelastic seizures (GSs), focal impaired awareness seizures (FIASs), and focal to bilateral tonic–clonic seizures (FBTCSs) being most prevalent. ² Surgical treatment of HHs leads to 67.7%–93.8% seizure freedom. ³ , ⁴

Different seizure symptoms may be related to specific brain network alterations. GSs, the most common seizure type in children with HHs, begin in infancy for more than one third of patients ⁵ , ⁶ and could be linked to brain networks responsible for emotion regulation and motor control that are crucial for producing laughter. ⁷ , ⁸ , ⁹ FBTCSs may involve thalamic circuits, ¹⁰ and mesial temporal lobe structures may be involved in FIASs. ⁸ Understanding the underlying brain network involved in different seizure syndromes in patients with HHs may assist in selecting more precise treatment options.

As a commonly used preoperative assessment tool for refractory epilepsy, positron emission tomography (PET), along with magnetic resonance imaging (MRI), plays a crucial role in the diagnosis and presurgical evaluation of epilepsy caused by HHs. In recent years, there has been increasing recognition that HH‐related epilepsy involves not only the lesion itself but also a more extensive epileptic network beyond the hamartoma. Specifically, Scholly et al. proposed the presence of secondary epileptic foci outside the hamartoma in HHs, mainly involving the amygdala and hippocampus. ¹¹ We also observed that some HH patients exhibited hypometabolism in the mesial temporal lobe, neocortex, and thalamus on the side of the hamartoma stalk attachment. ¹² Additionally, through stereoelectroencephalographic (SEEG) recordings during the epileptic seizures in HH patients, we found that the anterior cingulate cortex and medial frontal lobe were also involved in the onset and propagation of the seizures. ⁸ , ¹³ However, due to the limited spatial distribution of SEEG implantation and the resolution limitations of PET, previous studies have struggled to comprehensively explore the epileptic network that HH‐related epilepsy may involve across the whole brain. On the other hand, it is also unclear whether different seizure symptoms in HH are associated with distinct epileptic networks. Therefore, further investigation and validation of whole‐brain voxelwise characteristics and metabolic patterns across different seizure symptoms are still needed to better understand the epileptic networks in this rare disease.

In this study, we plan to investigate the structural and metabolic characteristics in people with HHs and identify the intrinsic brain networks that may serve the key hubs underlying the syndromes.

2. MATERIALS AND METHODS

2.1. Patient characteristics

We retrospectively screened the surgical database from the Department of Neurosurgery, Xuanwu Hospital, between July 2015 and December 2022 for individuals with HHs and refractory epilepsy, as defined by the International League Against Epilepsy. ¹⁴ All patients underwent a comprehensive presurgical evaluation by a multidisciplinary team (MDT), including assessments of seizure semiology, video‐electroencephalography (EEG), and structural MRI. HH lesions were identified in the subthalamic area using T1‐weighted imaging (T1WI) and T2‐weighted fluid‐attenuated inversion recovery imaging by experienced neurosurgeons (P.W. and Y.S.) and radiologists (B. Cui and Z. Wang). Other lesions including tumors in this area were excluded. Due to the need for intravenous injection of radioactive tracers and considerations of patient cooperation and family consent, fluorodeoxyglucose (FDG)‐PET scanning was performed only in a subset of patients as part of the preoperative assessment. All patients underwent SEEG of hamartomas as part of the invasive evaluation, with at least three ictal onset SEEG recordings, all demonstrating a clear origin in the hamartomas.

Seizure types included GSs, FIASs, and FBTCSs. Patients were divided into two groups for further analysis according to whether they had FBTCSs. The frequency of seizures was confirmed according to the following criteria: (1) for adult patients or those with normal cognitive function, seizure frequency was determined by combining self‐reported seizure diaries with caregiver records to ensure more accurate recall; and (2) for pediatric patients or those with cognitive impairments, seizure frequency was solely based on caregiver reports, as these patients may not reliably report their seizures. This study was approved by the ethics committee of Xuanwu Hospital of Capital Medical University, Beijing, China and was conducted in accordance with ethical standards of the Helsinki Declaration, 1975 (revised in 2000) regarding human experimentation (No. LYS [2019] 097). The data were collected for clinical decision‐making and are also available for research under institutional ethical approval.

2.2. MRI data acquisition and preprocessing

A Siemens 3‐T magnetization‐prepared rapid gradient‐echo sequence (1 mm) was performed on each patient following a standard scan. The structural MRI parameters were as follows: voxel size = 1.0 × 1.0 × 1.0 mm³, field of view = 256 × 256 mm, repetition time = 2476 ms, time to echo = 2.7 ms, inversion time = .9 s. The location of the HH was categorized as nonlateralized, left‐side predominant, or right‐side predominant by the MDT using T1WI/T2WI. The nonlateralized group was then subdivided into left‐ and right‐side predominant groups according to the EEG recordings of the laterality of the epileptic discharge.

MRI data were preprocessed with MATLAB 2018b (MathWorks) and the computational anatomy toolbox in Statistical Parametric Mapping version 12 (SPM12). ¹⁵ For spatial normalization, magnetic resonance images were coregistered with the Montreal Neurological Institute–International Consortium for Brain Mapping Average Brain 152 Atlas. Coregistered magnetic resonance images were segmented into gray matter volume (GMV), white matter (WM) volume, and cerebrospinal fluid images. To ensure a homogeneous group, such that all patients had an ipsilateral seizure propagation path, MRI scans of HHs classified as left‐side dominant were left–right flipped for further data analysis. Differences in GMV between patients with HHs and healthy controls (HCs) were evaluated using a GMV analysis, with covariates including age, sex, and total intracranial volume (TIV).

2.3. FDG‐PET data acquisition, preprocessing, and analysis

PET images of patients with HHs were acquired using a United Imaging PET scanner at Xuanwu Hospital (United Imaging). The patients were instructed to fast for at least 6 h and rest in a semidark room with their eyes closed and ears unplugged for 30 min after an intravenous injection of 3.7 MBq/kg FDG. PET images were acquired 30–40 min after injection, providing 2.4‐mm slices with an isotropic spatial resolution of 5 mm. PET images of HCs were obtained from the Chinese PET database provided by Wang et al., ¹⁶ using a Biograph 64 PET/CT Scanner (Siemens) with a slice thickness of 1.5 mm, an image matrix size of 336 × 336 × 110, and a pixel size of 1.01821 mm. The PET data were analyzed using SPM12 in MATLAB 2018b. To minimize differences in imaging between subjects and scan sequences, we standardized all subjects using the spatial and intensity normalization method described by Wang et al. ¹⁶ This process involved spatially registering all PET images to the China 2020 template provided by Liang et al. ¹⁷ and adjusting for scan sequence variations by dividing the PET image intensity by the average of pixels within 40%–90% of the maximum image intensity. We then flipped left‐side dominant HH images to right‐side dominant and used SPM12 to compare changes in metabolic patterns in patients with HHs with different semiologies, using age and sex as covariates and whole brain as reference region.

2.4. Statistical analysis

Demographic data were analyzed using SPSS (version 20.0, SPSS Inc.). Demographic data between groups were compared using the independent samples t‐test or Mann–Whitney U‐test depending on the different data formats, with p < .05 as the prespecified threshold. For GMV and FDG‐PET data analysis, we compared the structural and metabolic differences between HH patients with and without FBTCSs and HCs using a two‐sample t‐test adjusting for age, sex, and TIV. To further examine the relationship between structural and metabolic differences and seizure severity in patients with HHs, we performed multiple regression analysis with seizure frequency and disease duration as dependent variables in SPM12. The range of the cluster was set to K > 5, and familywise error (FWE) rates were used for multiple test corrections.

3. RESULTS

3.1. Demographic and clinical

Fifty‐nine patients with epileptogenic HHs (35 right hemisphere onset, 19 females, age = 10.2 ± 8.3 years, epilepsy duration = 7.0 ± 7.0 years, seizure frequency of GSs, FIASs, and/or FBTCSs per month = 272 ± 1099) were included in our study (Table 1, Figure 1A). Twenty‐seven patients had PET scans (15 with FBTCSs and 12 without FBTCSs).

TABLE 1.

Clinical and demographic characteristics of study cohort.

Characteristic	FBTCS group	nFBTCS group	HCs (GMV)	p	FBTCS group	nFBTCS group	HCs (PET)	p
Total number	28	31	30	—	15	12	45	—
Age, years (median, IQR)	1.6–33 (11.0, 13.75)	2–27 (5.0, 8.0)	19–30 (24.5, 5.25)	<.01*	2–33 (15.1, 7.0)	2–27 (8.8, 13.0)	24–48 (40.0, 8.0)	.06
Sex
Male	19	21	15	>.999	10	9	31	.64
Female	9	10	15		5	3	14
Duration of epilepsy, years, median, IQR	7.0, 13.38	3.0, 6.0	—	.03*	9.8, 14.0	6.5, 5.75	—	.28
Lateralization
Left	11	13	—	>.999	7	4	—	.48
Right	17	18	—		8	8	—
Delalande
I	6	4	—	.10	3	3	—	.98
II	6	13	—		3	2	—
III	14	8	—		6	5	—
IV	2	6	—		3	2	—
Frequency of focal seizures, n/month	174 ± 568	359 ± 1425	—	.51	65 ± 168	158 ± 172	—	.15
TIV, cm3, mean ± SD	1456 ± 160.9	1417 ± 206.5	—	.42	1514 ± 149.9	1444 ± 138.6	—	—
HH volume, mm3, mean ± SD	1659 ± 1409	1860 ± 2308	—	.69	1981 ± 1529	1913 ± 1442	—	—

Note: Continuous variables are presented as mean ± SD. Independent unpaired t‐tests were used for continuous variables with Welch correction; chi‐squared tests were used for comparison of sex between groups. * indicates statistical significance.

Abbreviations: FBTCS, focal to bilateral tonic–clonic seizure; GMV, gray matter volume; HC, healthy control; HH, hypothalamic hamartoma; IQR, interquartile range; nFBTCS, non‐FBTCS; PET, positron emission tomography; TIV, total intracranial volume.

FIGURE 1.

(A) Correlation matrix of demographic data by correlation analysis. The age at onset of epilepsy had a significant close relation with duration of epilepsy and seizure frequency, whereas the hypothalamic hamartoma (HH) volume had no relationship with these features. (B) The violin plot of HH volume between the subgroups. There were no differences in HH volume between the focal to bilateral tonic–clonic seizure (FBTCS) group and non‐FBTCS (nFBTCS) group (independent samples t‐test, p > .1). (C) The longer the duration of epilepsy and more frequent the seizures, the larger was the gray matter volume (GMV) in ipsilateral amygdala, piriform, and part of hippocampus. (D) GMV volume was decreased in ipsilateral postcentral cortex and inferior frontal cortex with longer duration of epilepsy and higher seizure frequency. (E) In 31 HH patients without FBTCSs, greater GMV in ipsilateral amygdala was related to longer duration of epilepsy and higher seizure frequency. There was no correlation in the FBTCS group. L, left; NS, not significant; R, right. * indicates statistical significance.

Patients with HHs were significantly younger than HCs in both MRI (15 males, 15 females, age median = 24.5 years, interquartile range [IQR] = 5.25 years) and PET (31 males, 14 females, age median = 40.0 years, IQR = 8.0 years) data groups (Table 1; Mann–Whitney test, p < .01), with no significant differences observed between groups in terms of sex (Table 1).

Correlation analysis of the demographic data showed a significant positive relationship between age at onset, epilepsy duration, and seizure frequency (Figure 1A). The greater the age at onset among HH patients, the longer was duration of epilepsy and the greater was the seizure frequency.

Twenty‐eight patients with HHs had FBTCSs, and 31 had no history of FBTCSs. HH patients with FBTCSs were significantly older and had longer duration of epilepsy than those without FBTCSs (Table 1; independent samples t‐test, p < .05).

3.2. Differences in PET and GMV between HH patients and controls

Compared to controls, HH patients exhibited increased GMV in several regions, including the ipsilateral amygdala, piriform cortex, hypothalamic area where the HH was attached, and bilateral temporal cortices (Table S1, Figure 2). No significant reductions in GMV were observed in HH patients. In terms of metabolic abnormality, HH patients showed hypometabolism primarily in the ipsilateral middle temporal gyrus and the surrounding WM compared to controls (Figure 4, Table S3; p < .05, FWE‐corrected). No areas of hypermetabolism were detected in HH patients. To test the robustness of our PET findings and to minimize potential confounding effects related to early brain development, we performed a secondary analysis excluding patients younger than 6 years. The results remained largely consistent with the original analysis, demonstrating similar patterns of hypometabolism (Figure S1).

FIGURE 2.

Gray matter volume (GMV) changes in 59 hypothalamic hamartoma (HH) patients compared with normal controls. The significantly increased GMV was found mainly in the ipsilateral amygdala, piriform, hypothalamus area, and bilateral temporal cortex after familywise error correction with sex, age, and total intracranial volume as covariates. There was no decreased GMV in HH patients compared with the control group.

FIGURE 4.

Glucose metabolic uptake changes in hypothalamic hamartoma (HH) patients compared with normal controls. Hypometabolic areas were mainly focalized at the ipsilateral middle temporal gyrus and the surrounding white matter in the whole HH patient group and HH with focal to bilateral tonic–clonic seizures (FBTCSs) group after familywise error correction. L, left; nFBTCS, non‐FBTCS; R, right.

3.3. Symptom‐specific abnormality patterns based on the presence of FBTCSs

When comparing the FBTCS group to controls, GMV was found to be increased in the ipsilateral hypothalamic area where the hamartomas were attached (Figure 3; p < .05, FWE‐corrected). In contrast, when comparing the non‐FBTCS group to controls, GMV was increased in the ipsilateral amygdala (Table S1, Figure 3; p < .05, FWE‐corrected).

FIGURE 3.

(A) Gray matter volume (GMV) changes in hypothalamic hamartoma (HH) patients with focal to bilateral tonic–clonic seizures (FBTCSs). Significantly increased GMV was observed in the ipsilateral hypothalamic area where the hamartoma base was attached (red arrows), after familywise error correction. A zoomed‐in view of this region is shown to better illustrate the focal change. No regions of decreased GMV were found in this group compared to controls. (B) GMV changes in HH patients without FBTCSs. The ipsilateral amygdala was significantly increased in the non‐FBTCS (nFBTCS) group compared with normal controls. There were no decreased brain regions in the comparison between these two groups. L, left; R, right.

FBTCS patients exhibited hypometabolism in the ipsilateral middle temporal gyrus compared to controls, which was similar to the findings in the overall HH group (Figure 4, Table S3; p < .05, FWE‐corrected). However, there were no significant metabolic differences between the non‐FBTCS group and controls.

No significant GMV differences were observed between the HH patient groups with and without FBTCSs. Likewise, there were no significant differences in the hamartoma volumes between these two groups (Table 1, Figure 1B). Additionally, no significant differences in FDG uptake were found between the FBTCS and non‐FBTCS groups.

3.4. Effect of clinical variables

Multiple regression analysis showed that in HH patients, a longer epilepsy duration and higher seizure frequency were associated with increased GMV in the ipsilateral amygdala, piriform cortex, and hippocampus, as well as a decreased GMV in the ipsilateral inferior frontal gyrus and postcentral cortex (Table S2, Figure 1C). Similar findings were observed in patients without FBTCSs, whereas these changes were not present in patients with FBTCSs (Table S2, Figure 1E). Through multiple regression analysis, we did not find any metabolism abnormalities in HH patients associated with epilepsy duration or seizure frequency. Similarly, no significant results were observed in either the FBTCS or NFBTCS groups.

4. DISCUSSION

In this study, we investigated the structural and metabolic abnormalities in patients with HHs and different seizure types. Despite patients being younger than controls, we observed increased GMV in the ipsilateral amygdala, piriform cortex, and hypothalamic region where the HH was attached. Although the volume increases in the hypothalamus were driven by patients with FBTCSs, increased volumes of amygdala and piriform cortex correlated with seizure frequency and duration of epilepsy, suggesting a wider network of abnormality beyond HHs, but restricted to the mesial temporal structures in patients with GSs and FIASs only. These findings indicate that HH‐related epilepsy may involve network‐level abnormalities that extend beyond the lesion itself.

The amygdala and piriform cortex were recently reported to be key hubs in the epileptic network in mesial temporal lobe epilepsy. ¹⁸ , ¹⁹ , ²⁰ The amygdala processes emotions, which might explain its significant involvement in HH patients, who all had GSs. ⁷ , ²¹ , ²² , ²³ The connection between amygdala and HH may result from the projection of stria terminalis from the hypothalamus, which is key in brain disorders associated with autonomic, neuroendocrine, and behavioral responses. ²⁴ , ²⁵ Previous studies, including SEEG, suggested that the mesial temporal lobe is an epileptogenic structure extending beyond the HH lesion. ⁸ , ¹¹ , ²⁶ A functional MRI (fMRI) study found significantly increased connectivity between the HH and the left amygdala, parahippocampal gyrus, cingulate gyrus, and occipital–temporal gyrus using seed‐based connectivity analysis, which partially aligned with our GMV and PET results. ²⁷ Ictal EEG‐fMRI studies revealed ictal propagation after several seizure events in HH, which also highlighted a role of the mesial temporal lobe. ²⁸ , ²⁹ The blood oxygen level‐dependent activation extended from the HH to the left fornix, temporal lobe, and subsequently through the cingulate fasciculus to the left frontal lobe. ²⁹ These observations suggested that our structural findings in the mesial temporal lobe are within an ictal dynamic HH network. Our previous SEEG study identified potential hubs of the HH epileptic network, where the cortical evoked potentials to HH stimulation were mainly detected in the amygdala ipsilateral to the attachment of the HH in one of our patients who experienced FIASs and GSs. ⁸ The cortical‐evoked potential results were consistent with our GMV study, suggesting a role of the amygdala and piriform cortex in patients with HHs with FIASs and/or GSs. In our current study, the volume of the amygdala and piriform cortex was larger the longer the epilepsy duration and the higher the seizure frequency, indicating that these structural changes are involved in ictogenesis. This suggests that early treatment for HH patients is crucial, as these abnormalities tend to worsen over time. We propose that evaluating extralesional abnormalities during early treatment could be important, which may involve targeted SEEG implantation in the amygdala and piriform cortex to assess seizure‐related network involvement. Additionally, individualized imaging assessments, such as generating z‐score maps for PET uptake and GMV, can help identify distinct abnormalities. If these are detected, early surgical interventions—such as thermal coagulation, laser interstitial thermal therapy, or resection of the affected structures—might be necessary to improve overall treatment outcomes.

Hypometabolism was prominently midtemporal gyrus driven by patients with FBTCSs. FDG‐PET has been used previously to investigate cerebral changes in patients with HHs. ¹² , ³⁰ These patients typically have one or more hypometabolic areas in the hemisphere predominantly affected by interictal and ictal epileptiform discharges. ¹² The distribution of hypometabolism tends to align broadly with the cortical network suspected to be primarily involved during nongelastic seizures. ³⁰ In addition to gray matter changes, we also observed hypometabolism extending into the adjacent WM around the middle temporal gyrus. Whereas FDG‐PET studies have largely focused on cortical abnormalities, previous research in other neuropsychiatric disorders—such as schizophrenia and bipolar disorder—has reported WM metabolic alterations, indicating that such findings may carry biological significance. ³¹ Moreover, structural studies have reported increased WM density in HH patients with multiple seizure types, ³² suggesting that the WM may be involved in the epileptogenic network. In patients with FBTCSs, widespread seizure propagation might contribute to axonal degeneration or altered metabolic demand along WM tracts, potentially explaining the observed hypometabolism. Future studies using multimodal imaging such as diffusion imaging and SEEG are needed to further clarify the role of WM in HH‐related epilepsy. Although we found hypometabolism in temporal lobes, as in previous studies, cortical hypometabolic patterns varied greatly among patients. Further investigations combining PET and EEG/SEEG are needed to validate the correlation between cortical and subcortical dysfunction with electrophysiological abnormality dynamically. Previous studies have investigated the relationship between HHs, seizure types, and other clinical manifestations. ¹⁰ , ²³ , ³³ These studies showed that a longer duration of epilepsy was associated with the development of other seizure types beyond GSs, whereas a larger lesion size was associated with increased risk of cognitive impairment and precocious puberty. ¹⁰ This is consistent with our finding that patients with HHs and FBTCSs had longer duration of epilepsy.

Few studies have used GMV analysis to quantitively assess changes at the whole‐brain level. ¹³ , ²⁷ One study found intergroup differences in WM density between patients with different seizure types, with higher WM density observed in the left temporal and right temporal–occipital regions in HH patients with multiple seizure types other than GSs. Additionally, greater WM density was noted in the bilateral cerebellar regions in the multiple seizure group compared to the GSs‐only group. These findings suggest the potential involvement of brain network hubs and WM changes in relation to different seizure symptoms. ³² Despite patients being younger than HCs, we observed increased GMV of amygdala and piriform cortex in HH patients, especially in those with GSs and FIASs only, suggesting an epileptogenic network involving especially the mesial structures in HHs with GSs and FIASs only.

Previous studies investigating the pathophysiology of GSs showed that GSs might originate from the mesial temporal lobe in some cases, associated with emotions rather than GSs arising due to HHs. It is challenging to consider the differential diagnosis for laughter spells and distinguish between different causes. ⁷ , ⁹ , ¹¹ , ³⁴ Our study suggests that, for patients with FIASs and GSs, a more thorough evaluation may be necessary to prevent missing any independent seizure origins outside the hamartoma. An additional SEEG electrode placed toward the amygdala may be critical in assessing the complete epileptic network of the HH, rather than solely targeting the hamartoma for intervention. However, in our current cohort, we did not place electrodes specifically in the amygdala, highlighting an area for improvement in future studies. Further research is required to elucidate the underlying mechanisms of GSs and to optimize the approach to identifying independent seizure foci.

This study found a positive correlation between epilepsy onset age and disease duration, which may seem contrary to conventional findings. We suggest that this phenomenon may be related to the clinical presentation of HHs and historical differences in epilepsy diagnosis. Although nearly all HH patients experience GSs in early childhood, these seizures are often subtle and may go unnoticed, leading to delays in diagnosis. In older patients, epilepsy was traditionally recognized primarily by the presence of FBTCSs, with GSs not widely understood at the time. As a result, many were only diagnosed later, either during routine medical evaluations or after developing secondary FBTCSs. In contrast, younger patients benefit from improved medical awareness and early diagnosis, with more proactive screenings and recognition of GSs as an epileptic manifestation, leading to earlier treatment and a shorter disease duration. Therefore, the observed positive correlation between age at onset and disease duration likely reflects the influence of historical diagnostic practices rather than an inherent feature of HH itself.

Our retrospective study has limitations. First, patients in our cohort are generally younger than controls; even though we have already included age as a covariate in our regression analysis, it is still possible that age could have some influence on the results. For example, recent studies have shown that GMV in the human brain peaks during youth and follows a dynamic pattern of gradual decline with age. ³⁵ This natural developmental trajectory might raise concerns about whether the observed GMV increases in these younger patients are due to pathological changes or represent a physiological delay in the expected reduction. Although our correlation analysis revealed a positive correlation between the volume increase in the amygdala and piriform cortex and disease duration, confirming the association between this change and the disease, further validation using age‐matched controls or brain aging models may be necessary in future studies to strengthen these results. Second, all our patients had GSs, so we cannot ascertain whether the amygdala and piriform cortex are involved in the epileptogenic networks in patients with GSs only. Third, due to the retrospective nature of this study, the extralesional abnormalities observed (e.g., in the amygdala and piriform cortex) lack validation from electrophysiological data. Further SEEG validation will be necessary in future studies.

5. CONCLUSIONS

We report structural abnormalities in patients with HH‐related epilepsy beyond the lesions. Our findings suggest that in patients with HHs, especially those with GSs and FIASs only, the amygdala and piriform cortex may play a role as hubs in an HH epileptic network. Comprehensive presurgical planning is needed for these patients to help identify individuals with independent seizure origins beyond the HH.

AUTHOR CONTRIBUTIONS

Yihe Wang, Tao Feng, Fenglai Xiao, Penghu Wei, Yongzhi Shan, and Guoguang Zhao designed the study, drafted a significant proportion of the manuscript, and contributed to the acquisition and analysis of data. Tao Feng, Yanfeng Yang, Penghu Wei, Yongzhi Shan, and Guoguang Zhao contributed to data acquisition. Marine N. Fleury, Lawrence P. Binding, Davide Giampiccolo, Peter Taylor, Matthias J. Koepp, and John S. Duncan contributed to important revisions to and comments on the manuscript. All authors reviewed and approved the final manuscript.

FUNDING INFORMATION

This study was supported by the National Natural Science Foundation of China (82030037), STI2030‐Major Projects (2021ZD0201801), Beijing Municipal Science & Technology Commission (Z221100007422016, Z221100002722007), and Translational and Application Project of Brain‐Inspired and Network Neuroscience on Brain Disorders, Beijing Municipal Health Commission (11000023 T000002036286). M.N.F. is supported by the Epilepsy Society. L.P.B. was supported by Epilepsy Research UK (P1904).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

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 the Ethics Committee of Xuanwu Hospital, Capital Medical University (no. LYS [2019] 097). Written informed consent was obtained from all participants.

Supporting information

Table S1.

Table S2.

Table S3.

Figure S1.

Wang Y, Feng T, Xiao F, Yang Y, Fleury MN, Binding LP, et al. Distinct gray matter and metabolic characteristics in hypothalamic hamartoma network with different semiology. Epilepsia. 2025;66:2853–2863. 10.1111/epi.18438

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.