A New Path Forward: Examining Traumatic Brain Injury Randomized Trials With Clinical, Biomarkers, Imaging and Modifiers

In June 2025, the National Institutes of Health-National Institute of Neurological Disorders and Stroke Classification and Nomenclature Initiative published the first new clinical characterization for acute traumatic brain injury (TBI) in over 50 years.1 The new model, referred to as the “clinical, biomarkers, imaging and modifiers (CBI-M)” model provides a long overdue reconceptualization of TBI across the spectrum of severity, rather than somewhat arbitrarily binning patients into groups labeled “mild” (traditionally defined as a Glasgow Coma scale [GCS] score of 13-15 at presentation), “moderate” (GCS = 9-12) or “severe” (GCS = 3-8). This considerable advancement invites the opportunity to re-evaluate past evidence from randomized controlled trials (RCTs) in this field and identify how this new model can refine future RCT efforts for TBI patients. There are currently no pharmaceutical treatment options to improve functional recovery after TBI, as most phase 2 and all phase 3 clinical trials have failed2; one possible reason is the lack of precision in study design that a new characterization model for TBI provides. To inform this editorial, we conducted a literature search on the Web of Science database on June 4, 2025 for the terms “traumatic brain injury” OR “concussion” OR “acquired brain injury” and “randomized clinical trial” OR “randomized controlled trial” since January 1, 2000 which resulted in 1202 articles (list available on request). We included articles which reported efficacy or effectiveness of a treatment of TBI patients in the primary outcome measure of an RCT design. We excluded articles which did not include human subjects, were not an RCT design, were a study protocol for a future RCT, did not include TBI patients, did not evaluate an intervention/treatment of TBI, whose primary aim was to evaluate feasibility, safety, or provide preliminary evidence, or was a secondary analysis of an RCT. This resulted in a final list of 248 articles, or roughly 10 per year since 2000. Overall, 122 studies (49.2%) reported a positive treatment effect for the primary outcome measure. Approximately 11% of studies (n = 26) did not describe GCS or severity at all for the included participants. The intervention types which had the highest rates of reporting successful treatment effects were alternative (ie, chiropractic manipulation, acupuncture; 80%), surgical (71.4%), and psychological therapies (68.9%) while the lowest rates of reporting successful treatments were behavioral (38.6%), cognitive (43.2%), and pharmacological therapies (43.5%). One possible explanation for the discrepancy in success rates may be a lack of specificity for the intervention and its intended patient population. Over the past 20 years, our conceptualization of TBI has shifted considerably towards an appreciation for its heterogeneity, both in clinical presentation and recovery. No two patients have the exact same experience with TBI, even two patients with the exact same GCS score, which is related to differences in personal experience and health before the injury itself, the nature of the TBI, and the care received.1 RCT designs in this population must also shift to embrace this heterogeneity and use more precise inclusion and exclusion criteria for human subjects. Our focus should also shift toward evaluating the effect of a targeted intervention designed for that specific subpopulation, rather than a blanket therapy prescribed to a blanket TBI population. Within the context of the CBI-M model, one could easily predict a stronger treatment effect for a mood-stabilizing medication in patients presenting with a GCS score of 15 (“C”), glial fibrillary acidic protein level of 80 pg/mL (“B”) with negative computed tomography scan (“I”) and history of untreated depression (“M”) compared with a trial enrolling all patients with “mild” TBI (ie, GCS score of 13-15 at presentation). Using more objective assessments both for characterizing TBI but also predicting responsiveness to treatment would likely make a meaningful improvement in the specificity of future RCTs in this population. Oncology trials have been using objective biomarker assays to inform inclusion criteria for over 35 years3; a practice that is now common in the field. In so doing, the authors design the RCTs to maximize the impact of the treatment of patients most likely to benefit. The introduction of the CBI-M model allows for targeted enrollment based on specific clinical characteristics. For example, in the hospital setting with resources to test acute blood biomarkers, it may be prudent to use the established reference ranges for healthy adults for glial fibrillary acidic protein (97.5 percentile = 51-75 pg/mL) to inform exclusion criteria. Furthermore, expanding beyond simply including those with a “positive” head computed tomography scan (ie, those with evidence of intracranial pathology) it may be useful to target interventions to specific pathologies in different cortical locations (ie, contusion, subarachnoid hemorrhage, epidural hematoma).4 Embracing this personalized medicine approach could yield meaningful results for translating different therapies into clinical practice. Patients with TBI have been battling the perception that their injury is not treatable, especially when the TBI is on the “milder” end of the spectrum where surgical or physiological interventions are not common.5 This may be related to the lack of available care pathways for individuals diagnosed with TBI in the hospital or outpatient settings.5 It is our contention that these perceptions and paucity of available follow-up care options is, at least in part, due to limited published, level 1 evidence supporting treatment effects. The advent of the CBI-M model should be recognized as an opportunity for significant refinement in RCT design to discover and/or introduce viable treatments for future TBI patients.