Abstract
Heiskanen M, Banuelos I, Manninen E, Andrade P, Hämäläinen E, Puhakka N, Pitkänen A. Epilepsia. 2024;65(12):3703-16.
Commentary
Each year 69 million people worldwide suffer a traumatic brain injury (TBI). 1 TBI's long-term impact on brain function can be profoundly debilitating. This is particularly true for post-traumatic epilepsy (PTE), which develops in up to 50% of patients with severe TBI 2 and is associated with notably poor quality of life. 3 Critically, PTE typically develops months to years after injury, opening a wide therapeutic window for prophylactic intervention. 2 Anticipating which TBI patients will develop PTE would allow for stratification of patients for antiepileptogenesis treatments and improve the success of clinical trials.
Blood-based biomarkers are particularly attractive early indicators of PTE because they are amenable to longitudinal sampling at low cost and minimal harm to the patient. 4 In a recent paper, Heiskainen et al. were successful in using bloodborne phosphorylated neurofilament heavy chain (pNF-H) measured 2 days after TBI to distinguish rats that developed PTE from those that did not 6 months later. 4 If they translate to humans, these findings could identify patients at risk for developing PTE and pNF-H could serve as one of the first blood-based biomarkers for PTE with genuine clinical value.
Having flagged pNF-H as a marker of axonal injury, Heiskanen et al. tested its predictive power by measuring blood pNF-H 2 days after lateral fluid percussion injury in adult rats. The team performed month-long 24/7 video/EEG recordings 6 months after TBI to capture seizures, including those that occurred only rarely. The authors then correlated pNF-H levels 2 days postinjury with PTE-related outcome measures such as the number of spontaneous seizures and their severity 6 months post-TBI. Remarkably, blood pNF-H levels at day 2 distinguished rats with and without PTE 6 months later with a sensitivity of 75% and a specificity of 58%. Moreover, pNF-H levels at the day-2 timepoint were positively correlated with the frequency and clustering of seizures, suggesting that the marker has added predictive value for PTE severity.
While Heiskanen et al.'s study is not the first to relate a blood-based biomarker to the development of PTE, 5 it has the advantage of large sample sizes (>100 TBI rats) and comprehensive controls (naïve and sham-operated). The study's rigorous design makes pNF-H a promising predictive biomarker of PTE in a field desperate for leads. Critically, pNF-H's long-term predictive value manifested as early as 2 days postinjury, early enough for meaningful clinical intervention. That said, repeated chronological sampling would provide valuable information about the coevolution of pNF-H with features of PTE. While earlier risk assessment holds the greatest clinical value, patients may not always present at the hospital immediately following injury. Would pNF-H still hold prognostic value 1 week later? What about a few months later? Future studies would also benefit from incorporating additional variables like sex and the type of TBI, known to influence PTE risk. 6
Injury severity is one of the few well-established risk factors for PTE. 6 The authors find that blood pNF-H levels 2 days postinjury positively correlate with lesion severity as measured by magnetic resonance imaging (MRI) conducted in the few weeks postinjury, in addition to histology done postmortem after EEG recording. One of the first steps in TBI assessment in humans is computed tomography (CT), an imaging approach that provides information on lesion severity. 7 While blood-based biomarkers like pNF-H are a cheaper alternative, clinicians may still prioritize CT for diagnosis and injury evaluation. This raises a key question: does pNF-H add prognostic value beyond CT for PTE risk, or is imaging alone sufficient?
Regardless, pNF-H's relationship to lesion size may provide insight into mechanisms underlying PTE beyond its potential role as a prognostic marker. Expressed exclusively by neurons, pNF-H serves as a pertinent indicator of axonal damage rather than a more general feature of injury like inflammation. 4 Although biomarkers are not necessarily causally linked to their associated measure (here PTE), it is possible that damage to neurons is a key trigger for PTE. Notably, a 2021 study from the same research group found that thalamic damage seen by MRI could serve as a biomarker for PTE development in the same rat model of PTE. 7 This raises the question of how thalamic damage and pNF-H are linked and whether pNF-H release by thalamocortical axonal disruption would be the ultimate prognostic indicator. The relationship between pNF-H levels and neuronal injury would also encourage further study into whether milder forms of TBI or diffuse versus focal impacts are associated with altered risk for developing PTE due to variable degrees of axonal damage.
Because TBI is often accompanied by a constellation of comorbidities, 4 the authors also evaluated pNF-H's predictive power for cognitive impairment and motor deficits. They found that blood pNF-H levels 2 days postinjury correlated negatively with performance on a comprehensive neuromotor test run on the same day as blood collection and again 4 and 12 days later. However, cognitive performance on the Morris water maze task approximately 1 month postinjury appeared unrelated to day-2 levels of pNF-H. Taken together, these data suggest that pNF-H's value may actually lie in its specificity to PTE and to early motor deficits but not cognitive impairment in the subacute phase.
Nevertheless, behavioral performance metrics may enhance pNF-H's predictive value when integrated with other measures, even if they lack strong predictive power on their own. For example, pNF-H levels combined with cognitive performance plus thalamic or cortical damage on MRI may more reliably predict PTE severity than pNF-H levels alone. If this is the case, a combinatorial approach using various modalities (e.g., blood-based biomarkers, imaging and EEG) may prove to be the optimal method to generate a comprehensive risk assessment. Given that the study generated a wealth of EEG data, future work could also incorporate other EEG-based indicators of PTE such as subclinical nonconvulsive epileptiform spikes or sleep-wake disruptions into pNF-H-based predictive models for improved accuracy and reliability.8-10
In summary, Heiskanen et al.’s study provides a valuable contribution to a field still lacking reliable early diagnostic markers for PTE. While pNF-H's predictive value remains to be reproduced preclinically and tested in patient populations with diverse demographics, this work represents an important step toward identifying biomarkers that could one day enable early intervention, more effective clinical trials, and improved patient outcomes.
