Serum biomarker trajectory clusters predict functional outcome and quality of life for traumatic brain injury

Abstract

Serum brain-enriched biomarkers are increasingly employed in the clinical evaluation of traumatic brain injury (TBI) to assist with triage, neuroimaging decisions, and prognostication. However, the potential of temporal biomarker trajectories to inform disease monitoring and long-term outcomes remains underexplored. We aim to identify distinct biomarker trajectory (TRAJ) profiles in traumatic brain injury patients and to examine their associations with long-term clinical outcomes. The study included 373, CT-positive Intensive Care Unit (ICU) traumatic brain injury patients (256 with initial Glasgow Coma Scale 3-12) from the Collaborative European NeuroTrauma Effectiveness Research in TBI (CENTER-TBI) core study who had at least two serum samples collected between days 1 and 5 post-injury. Six biomarkers -glial fibrillary acidic protein, ubiquitin C-terminal hydrolase-L1, neurofilament light chain, Tau, S100B, and neuron-specific enolase- were analysed. Optimal cluster solutions were determined using a composite validation index derived from seven internal clustering metrics. Distinct high and low trajectory classes emerged for all biomarkers; each comprising at least 40% of the cohort for five of the biomarkers. Cross-biomarker concordance analysis identified composite high (n = 104) and low (n = 110) TRAJ profiles. Key metrics for evaluating patient outcomes include Glasgow Outcome Scale Extended (GOSE), mortality, and Quality of Life after Brain Injury Overall Scale (QoLIBRI-OS) at 3, 6, and 12 months as well as a prognostic incremental value analysis using a conventional prediction model: International Mission for Prognosis and Analysis of Clinical Trials in TBI (IMPACT). High TRAJ membership is strongly associated with poor functional recovery (GOSE 1-4 at 3-12 months; odds ratio (OR) 8.79 [95% confidence interval (CI): 4.56-16.97]-12.29 [95%CI: 6.19-24.40], P < 0.001) and increased 180-day mortality (OR (14.84 [95%CI: 5.56-39.64], P < 0.001). Conversely, low TRAJ membership predicted favourable recovery (GOSE 6-8 at 3-12 months; OR 7.42 [95%CI: 3.10-17.76]-10.83 [95%CI: 3.65-32.14], P < 0.001) and better quality of life (QoLIBRI-OS ≥52; OR 4.98 [95%CI: 1.92-12.89], P < 0.01). Compared to single day-1 biomarker measurements, trajectory-based profiles yielded larger effect sizes and provided incremental prognostic value when added to the IMPACT prediction model (ΔR² 9-17%, P < 0.05). Overall, repeated biomarker measurements across the acute phase yield superior prognostic accuracy relative to single timepoint assessments. These findings underscore the importance of integrating longitudinal biomarker monitoring into ICU-based traumatic brain injury care and suggest that temporal trajectory profiling may improve prognostic modelling and facilitate more precise patient stratification for both clinical management and interventional studies.

Keywords: blood-based biomarkers; ensemble cluster validation model; machine learning; traumatic brain injury prognosis.

Graphical Abstract

Figure 1

Study selection and demographic characteristics. (A) CONSORT diagram for study subject selection from CENTER-TBI. (B) Demographic and injury characteristics of the final subject stratum.

Figure 2

Visualization of cluster analysis outcome across the 6 TBI biomarkers. (A) Most optimal trajectory clustering groups: low TRAJ and high TRAJ clusters using longitudinal data (Days 1–5). (B) Most optimal Day 1 clustering groups, with resulting clusters plotted across Days 1–5 for comparison (D2-D5 data not included in clustering analysis). Violin plots show median, minimum and maximum values, with frequency indicated by width. Each datapoint represents the biomarker level value of an independent patient sample by day. Sample sizes of class memberships are indicated by the numbers in brackets. Note that there were significant differences in the median biomarker levels between high and low TRAJ groups for each biomarker across all time points based on the Mann–Whitney U-test (P < 0.001).

Figure 3

Mortality counts at 6 months for each trajectory cluster and odds ratio (OR) of mortality in the high trajectory group. (††† P < 0.001).

Figure 4

Incremental value analysis of TRAJ clustering over IMPACT prediction model (core and extended variants) for both GOSE (poor recovery) and mortality outcome. The first bar shows the univariate Nagelkerke R² of the biomarker, while the other two bars represent ΔR² values of the biomarker, representing the additional variance explained in combination with the IMPACT core and extended models. For each individual biomarker, N = 373 while N = 214 for the composite biomarker group.