Focal to bilateral tonic-clonic seizures (FBTCS) are a feature of focal epilepsy that can lead to an increased risk of cardiac arrhythmias (1), seizure-related injuries (2), and sudden unexpected death in epilepsy (SUDEP) (3). Unlike other focal epilepsies, the mechanisms underlying FBCTS, in particular, remain elusive. Therefore, the importance of delineating the mechanisms involved in FBTCS is amplified as a vital objective to aid the control of FBTCS and the prevention of SUDEP.

Although the taxonomy of FBTCS implies whole-brain generalization of seizures, FBTCS are primarily highly selective, producing more vigorous activity in specific brain regions (4–6). Mounting evidence has endorsed subcortical structures such as the thalamus and basal ganglia and their associated circuits as critical to the information relay involved in FBTCS. The topographical arrangement of the thalamocortical relay fibres allow the projection to the cerebral cortex, from which sensory information is processed and relayed back to the original projection site in the thalamus. Acting as a “relay station” (7), the thalamus has widespread connections across the entire cerebral cortex (8) and moderates communication between various brain regions. Within the context of FBTCS, the thalamus has been proposed as a support system for seizure propagation via its role in the synchronization of abnormal cortical-subcortical ictal discharge (9, 10).

Additionally, the basal ganglia functions as a “braking system”, interacting with the thalamus and cortex through multiple parallel circuits, including the direct and indirect pathways (11). The basal ganglia are increasingly hypothesized to play an anticonvulsive role in FBTCS (12), yet the specific mechanisms remain unclear. Increased basal ganglia activity was reported to be negatively associated with FBTCS in TLE (13, 14). In contrast, others illustrated that the basal ganglia only become involved when ictal activity disperses to additional cortical regions during secondary generalization (15).

Tensor-based metrics derived from diffusion tensor imaging have been established as a valuable technique to quantify white matter changes in a range of epilepsies (16, 17). A meta-analysis showed patients with focal epilepsy had elevated regional mean diffusivity relative to controls (17). However, findings of altered fractional anisotropy (FA) are less consistent; some studies report reduced FA in patients with epilepsy compared to controls (18, 19), whilst others report no change across several regions (20, 21); thus, further clarification is required.

Moreover, the majority of diffusion MRI studies focus on TLE without the FBTCS subtype (17). One study that examined structural network alterations in patients with TLE alone vs. TLE with FBTCS used a “node abnormality” metric to quantify patient specific node abnormality load (22). They showed that patients with FBTCS in TLE had increased abnormality load in subcortical regions. However, their network approach did not allow the identification of specific tracts that may be involved in FBTCS. Though FBTCS can present under the umbrella of TLE, a reproducible and reliable method to identify specific tracts that may be involved in FBTCS would yield great clinical utility.

Advances in image analysis techniques, such as tract-weighted tensor-metrics, may improve the reliability of findings and enhance the interpretation of standard tensor-based metrics. A recent advancement to traditional tensor-based measurements is the introduction of track-weighted (TW) measurements. Here, tensor-based images are weighted by reconstructed streamlines traversing a given voxel to generate a TW version of the tensor-based image (23). TW measurements have been shown to improve reproducibility (i.e., TW-FA has significantly higher reproducibility than FA) and, for a given effect size, have increased power to detect group differences (23). Therefore, TW tensor-based measurements (termed “TW-TM” for brevity) could reveal subtle differences in epilepsy subtypes that would typically be undetected if quantified using only traditional tensor-based metrics.

Since the subcortical structures involved in FBTCS are also involved in focal seizures (24), we included both subtypes (focal only and focal with FBTCS) in this study to uncover distinct structural differences between the two groups. In addition to the thalamic and striatal regions, we included other white matter regions based on their well-documented role in focal seizures and functional connectivity (18, 25), to determine whether any observed group differences were unique to the FBTCS group, and to account for possible whole brain differences.

The innovation in this study is the combination of a neural network based tract segmentation model [Tractseg; (26)], with subject specific, TW-TM, to investigate specific tracts as potential biomarkers of seizure propagation in FBTCS. We hypothesized that compared to controls, patients with focal epilepsy would have significantly altered white matter in the subcortical regions, as observed through the TW-TM. We further hypothesized that compared to those without FBTCS, patients with FBTCS would have significantly altered white matter in the thalamic and striatal regions, as observed through the TW-TM.

In this study, tract-specific TW-TM were obtained to quantify the white matter differences in patients with and without FBTCS, compared to controls.

Our study revealed patients with FBTCS had unexpected, marked alterations in specific subcortical tracts considered critical to seizure genesis. Our findings indicated that the type of inferred white matter disruption (i.e., altered TW-AD) taken in the context of the subcortical region, may be relevant to the seizure semiology observed in FBTCS. The region-specific alterations are likely a function of seizure pathology rather than whole-brain differences. The ensuing discussion focuses solely on the compelling results that could yield clinical utility. Of particular interest were the whole brain tracts, where significantly different mean TW-TM were observed between the patient subgroups. Readers may refer to the Supplementary Tables to further examine other results.

The first remarkable finding was the significant difference between the two patient subgroups in the TW-ADC and TW-AD of the inferior fronto-occipital fascicle (IFO), and striatal and thalamic occipital tracts (STOCC and TOCC), providing support for the second hypothesis. The thalamocortical circuit has been implicated as a critical mechanism for the genesis of FBTCS (43) by engagement of the striatum and corpus callosum. In patients with mesio-TLE, stereotactic EEG (SEEG) has shown the role of thalamocortical connections in seizure termination (44). Altered thalamic functional profiles have also been proposed as imaging biomarkers of active secondary generalization (45).

Although the inhibitory role of the basal ganglia has been demonstrated in TLE with FBTCS, our study shows it may be important in other focal epilepsies presenting with FBTCS. Mounting research using stereo-EEG, deep brain stimulation, and EEG/fMRI suggests the physiological rhythms in the basal ganglia as a “pacemaker” for ictal discharge throughout adjacent regions (12). Stereo-EEG demonstrated cortico-striatal synchronization (46), implicating the changing synchronization as a mechanism to control the duration of abnormal oscillations within the striatal-thalamo-cortical loop and for potential termination. Deep brain stimulation studies show that regulation of basal ganglia activity via thalamostriatal projections may mediate generalized seizures (47). Combined EEG/fMRI measurements showed the thalamocortical-striatal network could be involved in activation (thalamocortical), deactivation (striatal), and consequently termination of cortical discharge (48). Furthermore, the cortico-striato-thalamo-cerebellar network has been implicated as a prominent feature of FBTCS via increased connectivity of structural covariance in the striatum and thalamus (49). Here, we identified alterations in the TW-TM of specific tracts (inferior fronto-occipital fascicle, striato-occipital, and thalamo-occipital tracts) that may represent a predictive biomarker to differentiate between individuals who develop FBTCS vs. those who do not. The increased TW-ADC in conjunction with the increased TW-AD could infer white matter disruption significant enough to inhibit propagation of ictal discharge. Our finding in a new dataset using reproducible imaging techniques provides the field with further evidence congruent with existing research. This finding adds to the foundational knowledge of FBTCS and could guide researchers in investigating the specific circuits and networks involved in FBTCS.

Second, the significant difference observed between the FBTCS-N and control groups in three TW-TM (TW-ADC, TW-FA, and TW-RD) derived from the thalamo-prefrontal and thalamo-premotor tracts might be explained by premorbid connections in the subcortical structures. A recent application of magnetic resonance fingerprinting in normal appearing MRI scans from a focal epilepsy cohort revealed predominantly ipsilateral changes in the tissue properties of the thalamus and basal ganglia (50), suggesting both the impairment and vulnerability of subcortical structures in focal epilepsy. In TLE, propagation patterns of electrophysiological discharge are bolstered by the structural white matter pathways which reinforce ictal propagation (51), suggesting that premorbid white matter damage may halt the natural progression of ictal discharge. However, for those with less altered TPREF and TPREM tracts, such as in our FBTCS-Y group, the white matter fibre bundles may bear a closer resemblance to the control group. Here, ictal discharge events could gain enough momentum to override the inhibitory mechanisms and spread to the contralateral hemisphere.

Lastly, given the overall significant differences were observed between the FBTCS-N group and controls, the significantly higher TW-AD of the middle cerebellar peduncle in the FBTCS-Y group is noteworthy. Unlike the thalamic tracts, the middle cerebellar peduncle is cross-hemispheric, connecting the cerebellum to the pons, and is composed entirely of centripetal fibres. Previous work has shown the role of the middle cerebellar peduncle in generalized tonic-clonic seizures (52), potentiating it as a tract where injury may increase susceptibility to the development of a generalized seizure.

Although increased AD has been shown to result from white matter maturation, all groups were age-matched, suggesting the possibility that the higher TW-AD in the middle cerebellar peduncle of the FBTCS-Y group may be due to distinct premorbid FBTCS mechanisms. Since increased AD can also imply better organization of fibre structure, the TW-AD differences between the FBTCS-Y and FBTCS-N groups in the cerebellar region could be a feature of the FBTCS brain whereby those with FBTCS-Y have stronger premorbid fibre connections and the ability to sustain the intrahemispheric flow of ictal discharge. The middle cerebellar peduncle is presented as a compelling region of interest in post-surgical imaging, and future works could investigate this further.

It is noted that although the inferior cerebellar peduncle makes up the peduncle group and is anatomically adjacent to the superior and middle cerebellar peduncles, it is neither functionally important in information relay nor inhibition, which may explain the absence of significant, between groups difference in this tract. Finally, there was no significant relationship between lesion presence and seizure onset side, or lesion presence and FBTCS, reinforcing the conjecture that the tract-specific TW-TM differences between the patient subgroups were due to premorbid connections rather than the effects of seizure injury. Our findings promote the possibility that propagation of ictal discharge from the epileptogenic zone to the contralateral hemisphere could be inhibited by premorbid damage to critical information relay and anatomically relevant tracts.

Imaging studies of patients with epilepsy are traditionally limited by small sample sizes (19–21, 53), primarily due to challenges in recruitment. Our relatively small sample size was a limitation in our study, which prevented further patient subgroup analysis of the striatal and thalamus tracts. Nevertheless, our study presents to the field specific tracts that may play a role in FBTCS, reinforcing the value of imaging techniques such as TW-TM in demystifying the mechanisms involved in FBTCS. Though generalized seizures may appear similar in semiology to FBTCS, specific pathways and networks may be involved in FBTCS once the bi-hemispheric ictal propagation begins and is a critical topic for further research. There is the possibility that the seizure onset zone may have a role in the observed differences. However, in the current sample size, such analysis would be low powered. Future work with a larger sample size would be better equipped to investigate such queries.

In summary, we used a robust and reproducible model to obtain track-specific TW-TM to show region-specific white matter alterations in a subgroup of patients with FBTCS, compared to those with focal epilepsy alone, and control participants. Our findings provide mechanistic insights into how structural changes may impact the functional role of the thalamic and striatal regions in individuals with FBCTS. We highlight specific tracts (IFO, STOCC, TPREF, and TPREM) that may be involved in FBTCS. Our results lay the foundation for a better understanding of seizure propagation in FBCTS and offer potential biomarkers that can help explain disease progression and aid treatment.

The datasets generated and/or analyzed in the current study are not publicly available as they are from RPAH patients, and only individuals named on the approved ethics are authorized to access the data. However, de-identified data can be made available on reasonable request to the corresponding author, subject to approval from the relevant governing ethics entities.

The studies involving human participants were reviewed and approved by the RPAH Sydney Local Health District (RPAH-SLHD) Ethics Committee. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

CM: conceptualization, methodology, data curation, imaging analysis, statistical analysis, manuscript writing, and review and editing. AD'S: imaging and diffusion analysis pipeline implementation, data curation and analysis, and manuscript revision. RZ: imaging pipeline—data acquisition and Tractseg implementation. MB and OK: conceptualization and manuscript revision. AN: clinical data acquisition, clinical advisory, conceptualization, methodology, and manuscript revision. CW: conceptualization, methodology, data curation, imaging analysis, statistical analysis, and manuscript revision. All authors contributed to the article and approved the submitted version.

This study received funding from UCB. The funder was not involved in the study design, collection, analysis, interpretation of data, the writing of this article or the decision to submit it for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.