Brain Age Index from Healthy Tissue Predicts Glioma-Related Epilepsy

The Challenge of Predicting Glioma-Related Epilepsy

Glioma-related epilepsy is a frequent complication that significantly impairs patient quality of life and complicates neuro-oncological management [1, 2]. While clinical features and molecular biomarkers, such as isocitrate dehydrogenase 1 (IDH1) mutations, provide some insight into seizure risk, accurately predicting which patients will develop epilepsy remains difficult due to complex tumor-host interactions [3, 4]. Standard neuroimaging traditionally focuses on the focal tumor burden, often overlooking the widespread structural and functional alterations that extend into seemingly healthy brain networks [5]. Consequently, clinicians lack reliable, non-invasive tools to assess the global cerebral impact of a localized tumor. A recent multicenter study offers a method to quantify these systemic brain changes, providing a quantitative neuroimaging metric to help predict epilepsy risk and guide early intervention in this vulnerable patient population.

Bypassing the Tumor to Measure Brain Age

Glioma frequently induces widespread structural and functional alterations extending far beyond the primary tumor site. Because of these extensive network disruptions, epilepsy is one of the most common clinical manifestations of the disease. While researchers often use magnetic resonance imaging (MRI) to estimate biological brain aging in psychiatric or neurodegenerative conditions, conventional brain-age models are rarely applied to neurosurgical diseases. This limitation occurs because focal structural damage violates the assumption of global anatomical integrity. When a tumor mass distorts the surrounding architecture, traditional whole-brain algorithms cannot generate accurate biological age estimates. To bypass the localized distortion caused by the lesion, the researchers proposed a Brain Age Index computed exclusively from non-tumorous brain regions. This metric integrates bias-corrected brain-age estimations with chronological-age normalization (a statistical adjustment that aligns the predicted biological age of the brain with the patient's actual calendar age to eliminate systematic calculation errors). By focusing solely on structurally intact tissue, the Brain Age Index was derived to quantify systemic cerebral alterations. For practicing physicians, this metric provides a way to measure the broader, brain-wide impact of a localized tumor, capturing the diffuse structural decline that contributes to seizure generation.

Multicenter MRI Data and Model Accuracy

To develop and validate this metric, the study utilized T1-weighted MRI data from 307 glioma patients across three centers. For comparison and baseline establishment, the researchers also included T1-weighted MRI data from 671 healthy controls. This large, multicenter dataset provided the necessary anatomical diversity to train an algorithm capable of recognizing typical brain aging patterns across different patient populations. Using these imaging scans, the researchers trained a residual convolutional neural network model for brain-age prediction. A residual convolutional neural network is a type of deep learning architecture specifically designed for complex image analysis, allowing the system to learn intricate structural patterns without losing data across multiple processing layers. When tested, the brain-age prediction model achieved a mean absolute error of 3.35 ± 4.19 years. This narrow margin of error indicates that the algorithm can reliably estimate a patient's biological brain age based solely on the structural appearance of healthy tissue, providing clinicians with a precise baseline to measure tumor-induced systemic alterations.

Adaptive Reorganization in Epileptic Patients

When applying the new metric to the study population, the researchers found that glioma patients exhibited significantly higher Brain Age Index values than healthy controls (p < 0.001). This finding indicates that the presence of a glioma accelerates the biological aging of structurally intact brain tissue far beyond normal chronological aging. However, a sub-analysis of the patient cohort revealed a counterintuitive pattern regarding seizure activity. Specifically, patients with glioma-related epilepsy showed reduced brain-age acceleration compared with non-epileptic patients. The authors note that this blunted aging effect in the seizure cohort provides insight into how the brain responds to a growing lesion. The reduced brain-age acceleration in patients with glioma-related epilepsy suggests possible adaptive neural reorganization. In clinical terms, adaptive neural reorganization refers to the brain's attempt to rewire its functional networks to compensate for tumor-induced damage. While this compensatory rewiring may help preserve overall structural integrity and slow the biological aging metric, the resulting abnormal network connectivity likely contributes to the generation of seizures.

Enhancing Predictive Models for Clinical Use

To translate these structural findings into a practical clinical tool, the researchers evaluated how well the new metric could forecast seizure risk. They developed a predictive algorithm that merged standard clinical data with radiomic features (quantitative data points extracted from medical images that are invisible to the naked eye). When tested, a combined clinic-radiomic model incorporating the Brain Age Index achieved an Area Under Curve (AUC) of 0.79 for epilepsy prediction. An AUC of 0.79 indicates a strong ability to distinguish between glioma patients who will develop epilepsy and those who will not, offering physicians a more objective method to stratify patient risk before seizures occur. Beyond seizure prediction, the findings establish the Brain Age Index as an imaging biomarker for detecting tumor-related cerebral alterations. By quantifying the diffuse impact of a localized lesion on the rest of the brain, this metric provides a measurable indicator of systemic neurological decline. Consequently, the Brain Age Index enhances prognostic modeling and functional network assessment in glioma. For practicing neuro-oncologists and neurologists, integrating this index into routine imaging analysis could improve the accuracy of patient prognoses and help map the functional integrity of brain networks, ultimately guiding more personalized anti-epileptic and oncological treatment strategies.

References

1. Ferreira MP, Carvalho RL, Borges DF, Soares JI, Casalta-Lopes J. The prevalence of post-therapy epilepsy in patients treated for high-grade glial tumors: a systematic review and meta-analysis.. Medical oncology (Northwood, London, England). 2025. doi:10.1007/s12032-025-02677-6

2. Avila EK, Tobochnik S, Inati SK, et al. Brain tumor-related epilepsy management: A Society for Neuro-oncology (SNO) consensus review on current management.. Neuro-oncology. 2024. doi:10.1093/neuonc/noad154

3. Hajikarimloo B, Mohammadzadeh I, Shirzadi P, et al. Machine learning-based models for predicting glioma-associated epilepsy: a systematic review and meta-analysis.. Discover oncology. 2025. doi:10.1007/s12672-025-04035-4

4. Li Y, Shan X, Wu Z, Wang Y, Ling M, Fan X. IDH1 mutation is associated with a higher preoperative seizure incidence in low-grade glioma: A systematic review and meta-analysis.. Seizure. 2018. doi:10.1016/j.seizure.2018.01.011

5. Liang S, Fan X, Kuang S, et al. Clinical practice guidelines for the diagnosis and treatment of diffuse glioma-related epilepsy: 2025 update.. Cancer letters. 2026. doi:10.1016/j.canlet.2026.218360