The researchers analysed data from the Concussion REcovery STudy, a prospective observational cohort in Perth, Western Australia. Thirty-five adults aged 18 to 65 with a medically diagnosed mTBI were included, alongside 34 age- and sex-matched controls without an mTBI in the previous five years. The injuries were uncomplicated: participants had no acute trauma-related abnormality visible on conventional CT or MRI, had less than 24 hours of post-traumatic amnesia and less than 30 minutes of loss of consciousness. Blood was collected, on average, about five and a half days after injury, and MRI was performed at roughly seven days. Participants then reported their symptoms by telephone at three, six and 12 months.

Blood testing measured several proteins associated with neural injury. NfL is a structural component of axons, the long projections that carry signals between nerve cells. Mechanical forces can stretch axons and disturb their internal scaffolding, allowing NfL fragments to escape into surrounding fluid and eventually enter the bloodstream. The team also measured brain-derived tau, glial fibrillary acidic protein and ubiquitin carboxyl-terminal hydrolase-L1, or UCH-L1, as well as the less commonly studied 4R-tau isoform. UCH-L1 was excluded because the assay results were too variable. Brain-derived tau and GFAP did not differ between groups, possibly in part because those proteins change rapidly after injury and may have returned closer to baseline by the time samples were collected.

NfL and 4R-tau were both higher shortly after mTBI than in controls, although their relationships with later symptoms pointed in opposite directions. Higher NfL was associated with a lower likelihood of still being symptomatic at three months. In the study’s statistical model, NfL correctly classified about three-quarters of participants according to three-month recovery status, but this estimate came from a small sample and does not establish a reliable prediction rule. By contrast, higher 4R-tau was associated with a greater likelihood of symptoms at 12 months and correctly classified about 74 per cent of participants in that analysis. The researchers stress that these results need replication before either marker could guide individual care. The apparently favourable association between higher NfL and recovery may seem counterintuitive. NfL is often treated as a sign of axonal damage, and previous studies have linked higher concentrations with worse outcomes. The authors propose that, in uncomplicated mTBI, a stronger but temporary biological response could sometimes reflect structural remodelling rather than irreversible degeneration. A rise in NfL might therefore accompany a transient disturbance that resolves. But the study collected blood at only one acute timepoint, so it could not determine whether each person’s NfL was rising, peaking or already declining. The result could also reflect chance, unmeasured differences between participants or the small number of people in each recovery group.

4R-tau offers a different biological possibility. Tau proteins help stabilise microtubules, the internal tracks that support axonal transport. The four-repeat form binds particularly strongly to tubulin, but when it detaches it may contribute to microtubule instability, interfere with mitochondrial transport and disrupt calcium regulation. The new study cannot show that 4R-tau caused persistent symptoms. Nevertheless, its selective elevation in participants who remained symptomatic—without a corresponding rise in brain-derived tau—raises the possibility that particular tau isoforms capture processes that total tau measurements miss. The researchers say future studies should directly measure the balance between three-repeat and four-repeat tau, and follow the markers over time.

The MRI results supplied a complementary view of the injury. Diffusion MRI tracks the movement of water molecules through tissue, allowing researchers to infer changes in white-matter organisation that conventional scans may miss. Fractional anisotropy reflects how strongly diffusion follows one direction, as it tends to do along coherent nerve fibres. Mean diffusivity captures the average magnitude of diffusion, while axial and radial diffusivity describe movement along and across the principal fibre direction. These measures are sensitive to microstructural change, but they are not biologically specific: a change can reflect fluid shifts, inflammation, altered cell structure or other processes rather than one identifiable lesion.

Across 48 white-matter tracts, people with mTBI showed widespread increases in radial and axial diffusivity, together with more limited changes in mean diffusivity and fractional anisotropy. The patterns were especially evident in long projection and association fibres, including regions such as the anterior corona radiata, cingulum and superior longitudinal fasciculus. Participants who remained symptomatic at three months had particularly high radial diffusivity in several tracts. Because radial diffusivity is sometimes associated with changes perpendicular to the fibre direction, it can be interpreted as a possible indicator of myelin disruption. In the acute setting, however, the authors consider disturbed ion balance, extracellular fluid and inflammatory processes more plausible than overt demyelination. Mean diffusivity showed a contrasting pattern. Participants who later recovered had higher acute mean diffusivity than controls across several association-fibre regions, and higher mean diffusivity in the left superior longitudinal fasciculus was associated with a lower likelihood of symptoms at six months. The relationship persisted in a model of 12-month status, although the wording of the statistical result indicates an inverse association rather than a proven protective effect. Recovered participants also showed relationships between NfL and mean diffusivity in the right cerebral peduncle, while lower 4R-tau was associated with higher axial diffusivity in the right posterior thalamic radiation among those recovered at 12 months. No corrected blood–MRI correlations emerged in the symptomatic group, perhaps reflecting greater biological heterogeneity or processes outside white matter.

The study also examined whether resilience and coping style changed the links between acute biology and later symptoms. A general measure of resilience did not distinguish injured participants from controls and did not significantly moderate the principal biomarker relationships. Coping patterns did. Passive reaction coping significantly altered the association between NfL and three-month outcome: the inverse relationship between NfL and persistent symptoms was strongest among participants reporting more passive coping. Reassuring-thought coping similarly moderated the relationship between 4R-tau and 12-month symptoms; higher 4R-tau was associated with greater odds of persistent symptoms when this coping style was less frequently used. These findings do not mean that a particular coping strategy causes recovery or that people can think their way out of brain injury. Coping may reflect premorbid traits, stress biology, illness perceptions or the effects of early symptoms, and the analyses were explicitly hypothesis-generating.

Several limitations make the results a starting point rather than a diagnostic breakthrough. The cohort was small, with follow-up numbers falling to 31 participants at six and 12 months. Participants were classified as recovered or symptomatic using self-reported Post-Concussion Symptom Scale scores, so the outcome necessarily reflected subjective experience; symptoms also occur in the general population. Recovery-group membership could change from one follow-up to another, making long-term trajectories difficult to interpret. The study used a single acute blood and MRI assessment, lacked pre-injury measurements of resilience and coping, and included multiple statistical comparisons despite limited power. Diffusion metrics cannot by themselves identify the underlying tissue process, and the blood assays for 4R-tau and brain-derived tau are not directly interchangeable.

Even with those caveats, the results point toward a more nuanced way to study mTBI. Instead of treating every concussion as biologically equivalent, researchers could combine clinical assessment with molecular markers, microstructural imaging and personal context. The next tests will require larger prospective cohorts, repeated blood sampling and serial diffusion MRI, ideally with pre-injury data in groups such as athletes at elevated risk. Studies should determine whether NfL and 4R-tau follow distinct time courses, whether tau-isoform balance predicts symptoms independently of initial symptom burden, and whether the MRI patterns replicate across different injury mechanisms and healthcare settings. For now, the central message is not that a blood test can forecast an individual’s future, but that the early biology of mTBI may contain several different stories—some associated with resolution, others with symptoms that persist.

Subject of Research: Mild traumatic brain injury, blood-based biomarkers, diffusion MRI and recovery trajectories.

Article Title: Divergent recovery trajectories after mild traumatic brain injury are characterized by distinct acute profiles of neurofilament light, 4R-tau, and white matter diffusivity.

Article References: Original article in the Journal of Neurology.

DOI: 10.1007/s00415-026-14060-0

Keywords: Mild traumatic brain injury; post-concussion symptoms; blood biomarkers; diffusion magnetic resonance imaging; psychological resilience; coping skills.