Assessment of Blood Biomarker Profile After Acute Concussion During Combative Training Among US Military Cadets

Key Points Question Is concussion incurred during combative training associated with changes in blood levels of protein biomarkers for traumatic brain injury among military service academy cadets? Findings In this case-control study of 103 military service academy cadets in the US, the levels of glial fibrillary acidic protein and ubiquitin C-terminal hydrolase-L1 biomarkers among cadets who incurred concussions during combative training indicated significant increases compared with baseline values and values found in matched cadets who participated in the same combative training exercises but did not incur concussions. Neurofilament light chain levels were significantly higher among cadets with concussions compared with those without concussions at less than 6 hours after injury to 7 days after return to activity. Meaning In this study, blood biomarkers delineated physiological changes after concussions that were incurred during military combative training, indicating patterns that were largely consistent with those reported in studies of sport-associated concussions.


Introduction
Mild traumatic brain injuries (TBIs) or concussions are a substantial public health problem, occurring with peak incidence in adolescence and young adulthood. 1 Common mechanisms for concussion in this age range are typically associated with sports and other recreational activities. 2 Although several studies have examined acute injury response and recovery from sport-associated concussions, recreational and occupational injuries have been less characterized. Military service academy (MSA) cadets represent a group of young, healthy individuals who have a higher risk of incurring mild TBI in the setting of combative training. Cadets are required to participate in combative training exercises, which, in the context of this study, were defined to include boxing, self-defense, martial arts, and other military training, all of which may result in head impacts. It is essential that studies of combative training-associated concussions have an active control group of cadets who are undergoing the same combative training exercises but have not incurred concussions (defined as a contact-control group). Symptom burdens associated with the stress of combative training may overlap with those of concussion, and the potential for incurring subconcussive impacts as part of combative training needs to also be considered. 3 Data also indicate that aerobic activity in the absence of head injuries has implications for blood biomarkers of concussion. 4,5 Blood biomarkers have been increasingly studied in military settings because acute clinical assessment of concussion may be challenging under the conditions of both military training and combat deployment. 6,7 Conflicting sources of symptoms (eg, pain and anxiety) can confound symptom-based concussion assessment tools. For cadets, concerns about the career implications of a concussion diagnosis may produce underreporting of symptoms, underdiagnosis, and premature return to activity and duty. Active-duty personnel's desire to return to their units may similarly complicate accurate concussion diagnosis and return-to-duty dispositions. Quantitative biological measures to inform concussion diagnosis have important implications in the military that may exceed their role in conventional civilian settings.
Blood-based biomarkers offer the potential for objective detection of molecules associated with the underlying pathophysiologic changes associated with concussion. 8 Glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) levels have been associated with positive computed tomographic findings in patients presenting to the emergency department with mild TBIs. 9,10 Higher levels of GFAP have also been reported in patients with mild TBI and positive magnetic resonance imaging findings. 11,12 Increased neurofilament light chain (NF-L) levels have been associated with TBI, concussion, and persistent postconcussive symptoms as well as other neurological disorders in the absence of TBI. 13,14 Increased tau levels have been implicated in repetitive mild TBI, but study results have been mixed. [15][16][17] This study targeted concussions that occurred in combative training among MSA cadets to assess whether blood-based biomarkers (GFAP, UCH-L1, NF-L, and tau) indicated patterns similar to those observed after sport-associated concussions or mild TBI. The overarching hypothesis was that combative training-associated concussions would be associated with increased brain injuryassociated proteins in the blood and that these biomarkers could enhance the accurate detection of concussion and recovery among MSA cadets.

Study Design and Population
This study was part of a larger multicenter prospective cohort study conducted by the Advanced A total of 1178 MSA cadets completed baseline clinical assessments, with 1085 cadets also providing a blood sample at baseline. Cadets who incurred concussions in the setting of combative training (n = 67; concussion group) were matched on a 2:1 ratio (based on institution, sex, and sport category) with cadets who participated in the same combative training exercises but did not incur concussions (n = 36; contact-control group). Sport categories were football, hockey, soccer, and other. Concussion was defined using the consensus definition from the US Department of Defense evidence-based guidelines. 20 Both groups underwent the same protocols for baseline and serial assessment, including clinical measures and blood sample collection.

Procedures, Collection, and Bioassays
The CARE Advanced Research Core protocol involved preseason baseline clinical testing and blood sample collection. The CARE Advanced Research Core postinjury protocol included collection of clinical data and blood samples from cadets with concussions at the following assessment points: (1) acute postinjury (<6 hours); (2) 24 to 48 hours postinjury; (3) asymptomatic postinjury defined as the point at which the cadet reported being asymptomatic and the return-to-activity (RTA) protocol was initiated; and (4) 7 days after unrestricted RTA (7-day post-RTA). Not all participants completed all visits.
Clinical data obtained at all postinjury assessment points included the Sport Concussion Assessment Tool-Third Edition (SCAT-3) symptom severity evaluation (symptom severity score range, 0-132, with higher scores indicating more severe symptoms), the Standardized Assessment of Concussion (SAC; score range, 0-30, with higher scores indicating better cognitive functioning), and the Balance Error Scoring System (BESS; score range, 0-60, with higher scores indicating worse static postural stability). The 18-item Brief Symptom Inventory (BSI-18; score range, 0-72, with higher scores indicating higher levels of psychological distress) was obtained at all assessment points with the exception of the acute postinjury visit.
Nonfasting blood samples were collected using standard venipuncture at baseline and at all postinjury assessment points in a 10-mL red-top tube for serum and allowed to clot for 30 minutes.
Samples were then centrifuged within 60 minutes of blood collection at 1500g for 15 minutes. Next, they were aliquoted into cryovials, frozen at −80°C, and shipped on dry ice to the CARE Consortium biorepository at the Indiana University School of Medicine National Institute of Neurological Disorders and Stroke BioSEND Core. Samples were then sent to the National Institutes of Health for analysis.
Biomarker levels were assayed using single-protein array technology (Simoa; Quanterix Corp) in simultaneous multiplex assays to quantify GFAP, UCH-L1, NF-L, and tau levels (pg/mL). Assays were batched to minimize variability, with each batch run with appropriate standards and controls to ensure reliability. Groups were distributed randomly across plates, and longitudinal samples from the same individual were run on the same plate to reduce potential batch effects. All samples were analyzed in duplicate, and the technicians running the assays were blinded to all data. In rare instances in which coefficients of variance exceeded 20%, samples were rerun. Data were not used if either intra-assay or interassay performance was higher than 20%, which occurred in less than 5% of samples. The lower limits of quantification were 0.241 pg/mL for NF-L, 0.467 pg/mL for GFAP, 5.45 pg/mL for UCH-L1, and 0.053 pg/mL for tau.

Statistical Analysis
Statistical analyses were performed using IBM SPSS Statistics, version 24.0 (IBM Corp). Demographic characteristics and clinical outcome group comparisons at baseline were evaluated using t tests and χ 2 or Fisher exact tests. Biomarker levels and estimated mean differences in biomarker levels were natural log (ln) transformed to decrease the skewness of their distributions. Linear mixed-effects models were used to evaluate changes in biomarker levels and to examine the results of clinical assessments (ie, SCAT-3, SAC, BESS, and BSI-18 scores) within cadets over time as a function of group, with visit modeled as a repeated factor (ie, baseline, acute postinjury, 24-48 hours postinjury, start of RTA protocol, and 7 days post-RTA), group (ie, concussion and contact-control), and group-by-visit interaction. Estimated mean differences and 95% CIs were provided for linear mixed-effects post hoc analyses. Bonferroni correction to protect the familywise error at α = .05 was conducted for post hoc tests, when indicated. Sensitivity analyses were performed to assess whether acute injury severity (based on the SCAT-3 symptom severity score)differed significantly between cadets in the concussion group who did and did not have blood samples collected at the acute postinjury point.
Pearson correlation coefficients were used to examine the association between biomarkers and clinical outcome measures.
Receiver operating characteristic curves and area under the curve (AUC) with 95% CIs were used to quantify the ability of markers to discriminate cadets with concussion and those without concussion at the acute and 24-to 48-hour postinjury points. Logistic regression analyses were conducted to examine whether the inclusion of blood biomarkers was significantly associated with greater discrimination of cadets in the concussion group vs the contact-control group compared with self-reported symptoms alone. Tests were 2-sided with a significance threshold of P < .05. Data were analyzed from March 1, 2019, to January 14, 2020.

Demographic Characteristics
Of the 1178 MSA cadets with full clinical baseline data and blood samples, 217 cadets incurred a concussion; of those, 210 cadets had blood samples collected after injury. A total of 77 out-of-sport concussions (defined as concussions incurred outside of varsity, organized, or intramural sports) were identified. Two cadets each had 2 concussions during the study period, and only the first concussion was included in our analysis. Out-of-sport concussions that were not associated with combative training were excluded (1 concussion was associated with snowboarding, 6 concussions were associated with unorganized sports, and the cause of 1 concussion was undetermined). These exclusions produced a final sample of 103 cadets; of those, 67 cadets were in the concussion group

JAMA Network Open | Neurology
Blood Biomarker Profile After Acute Concussion in Combative Training Among Military Cadets and 36 matched cadets were in the contact-control group. The mean (SD) age of cadets in the concussion group was 18.6 (1.3) years, and 40 cadets (59.7%) were male. The mean (SD) age of matched cadets in the contact-control group was 19.5 (1.3) years, and 25 cadets (69.4%) were male.
No statistically significant differences in sex, years of sport participation, number of previous concussions, race, or ethnicity were found between groups. Cadets in the concussion group were significantly younger (mean difference, −0.87; 95% CI, −1.40 to −0.34; P = .001), of shorter height (mean difference, −4.33; 95% CI, −8.19 to −0.47; P = .03), and of lower weight (mean difference, −6.57; 95% CI, −12.95 to −0.19; P = .04) than those in the contact-control group but did not significantly differ in body mass index (calculated as weight in kilograms divided by height in meters squared; mean difference, −0.92; 95% CI, −2.29 to 0.44; P = .18). Although there were no statistically significant group differences in SAC or BESS scores at baseline, cadets in the concussion group had higher baseline SCAT-3 symptom severity scores (mean difference, 6.89; 95% CI, 2.34-11.43; P = .003) and higher BSI-18 global severity scores (mean difference, 3.61; 95% CI, 1.19-6.04; P = .004) than those in the contact-control group. Descriptive statistics for clinical outcome assessments at baseline and all postinjury points are provided in eTable 1 in the Supplement.

Clinical Measures
For clinical measures, the main effects of group (concussion vs contact-control) were observed in the SCAT-3 symptom severity score (F 1,91.1 = 55.6; P < .001), the BESS score (

Primary Biomarker Analysis
Results of the linear mixed-effects analysis revealed a significant main effect of group for GFAP levels (F 1,102 = 9.06; P = .003) and a significant time by group interaction (F 4,283 = 6.00; P = .02).

Contact-control group
The concussion group includes cadets who incurred concussions during combative training, and the contact-control group includes matched cadets who participated in the same combative training but did not incur concussions. Acute indicates the acute postinjury (<6 hours) point; 24-48 h, the 24-to 48-hour postinjury point; asymptomatic, the postinjury point at which the cadet reported being asymptomatic and return-toactivity (RTA) protocol was initiated; and 7d post-RTA, 7 days after unrestricted return to activity. Error bars represent plus 1 and minus 1 SE. A, Symptom severity score range, 0 to 132, with higher scores indicating more severe symptoms. B, Score range, 0 to 60, with higher scores indicating worse static postural stability. C, Score range, 0 to 72, with higher scores indicating higher levels of psychological distress. D, Score range, 0 to 30, with higher scores indicating better cognitive functioning.

Receiver Operating Characteristic Analysis
To evaluate the ability of these biomarkers to distinguish cadets in the concussion group from those in the contact-control group, receiver operating characteristic curves and AUC analyses were conducted. Individual biomarkers were assessed at the acute and 24-to 48-hour postinjury points (Figure 3).
The GFAP and UCH-L1 biomarkers combined had an AUC of 0.78 (95% CI, 0.66-0.91; P < .001), and the AUC for all 4 biomarkers combined was 0.80 (95% CI, 0.68-0.93; P < .001) at the acute postinjury point. The AUC values were lower at the 24-to 48-hour postinjury point. Overall, the addition of blood biomarkers was not significantly associated with greater discrimination between the concussion and contact-control groups beyond the SCAT-3 results alone at either the acute or 24to 48-hour postinjury points.

Clinical Outcome Measures and Biomarkers
In addition, Pearson correlations were used to examine the association between biomarkers and clinical outcomes at the acute and 24-to 48-hour postinjury points (eTable 7 in the Supplement).
Measures were matched at the specific time (ie, acute biomarker with acute clinical outcome) and across times (ie, acute biomarker with 24-to 48-hour clinical outcome).
Overall, Pearson correlations with clinical outcomes were small to moderate (rՅ0.4 for all outcomes). Only the association between acute GFAP level and 24-to 48-hour BESS score was statistically significant among cadets in the concussion group (r = 0.36; P = .049).

Discussion
To our knowledge, this study is the first to examine combative training-associated concussions among MSA cadets using longitudinal measures of multiple protein biomarkers in blood. Notably, the control group for this investigation consisted of other cadets undergoing the same type of combative training exercises as those with concussions and thus constituted an active contact-control group, helping to account for the separate outcomes associated with combative training exercises for biomarker changes, including repetitive impacts without clinical symptoms. Although these findings are newly described in the cohort of combative training-associated concussions among military cadets, acute increases in GFAP and UCH-L1 levels have also been observed in individuals with sportassociated concussions 21 and mild to moderate TBI who presented to the emergency department. 10 These data suggest that GFAP and UCH-L1 biomarkers have potential for use in future studies of concussion in the combative training setting, with UCH-L1 having a relatively early period of detection and GFAP having longer-lasting increases.
After these injuries, MSA cadets experienced acute clinical changes, as measured by increased SCAT-3 symptom severity scores, increased BESS scores, and decreased SAC scores with recovery over time. The BSI-18 global severity index score was high at baseline and at the acute postinjury point, but it also indicated recovery with time. Notably, Pearson correlations between clinical measures and acute or 24-to 48-hour postinjury biomarkers were not robust. This finding suggests the potential utility of objective biomarkers to help identify injury in cadets with few symptoms or clinical signs.
Neurofilament light chain levels indicated persistent differences between cadets in the concussion and contact-control groups; these differences are interesting but distinct from previous studies of NF-L levels after concussion and suggest that NF-L levels may indicate ongoing axonal injury after TBI. In boxers, NF-L levels in cerebrospinal fluid have been found to peak at 15 days and normalize after 3 to 9 months. 22,23 Although no increases in NF-L levels were observed in a study of soccer players after a heading training session, 24 increased levels of NF-L were found in athletes playing collision sports, such as American football 25 and professional Swedish ice hockey. 14 This study provides additional support for NF-L as a subacute concussion biomarker, with potential value even days after injury.
Also notable is the finding that increases in GFAP and NF-L levels persisted beyond the point of  that, in addition to diagnosis, biomarkers may help to identify when an individual has recovered sufficiently to return to full duty and activity.

Limitations
This study has several limitations. The cohorts overall are relatively modest in size. Some group differences were present at baseline for SCAT-3 symptom severity and BSI-18 global severity. It is possible that the age difference between cadets in the concussion and contact-control groups partially explains this finding, with younger cadets in the contact-control group more likely than older cadets in the contact-control group to be experiencing the initial stress of the academy and basic training. These differences in baseline measures may have consequences for the association between clinical and biomarker findings; however, they would not necessarily interfere with the objective assay measures of individual biomarkers.
Another consideration is that the number of cadets in the concussion group who also had baseline blood sample collection for biomarkers was relatively small, producing higher baseline variance among those with concussions and potentially obscuring subtle within-group differences between baseline and the acute postinjury point. The sensitivity analysis indicated similar acute and 24-to 48-hour SCAT-3 symptom severity scores among cadets with and without blood samples at the acute postinjury point, reducing concern about bias among cadets for whom blood samples were available.
Overall, the receiver operating characteristic analysis indicated that the addition of blood biomarkers did not provide greater discrimination between cadets in the concussion and contactcontrol groups than did SCAT-3 scores at the acute or 24-to 48-hour postinjury points. However, the patterns of biomarker change reported after combative training concussions in the present study were consistent with those found in studies of concussion and TBI in other settings. 10,21 These patterns may better inform future work directed at translating these and other blood biomarker findings to develop the clinical end points (eg, positive and negative predictive values) that will be necessary to determine clinical utility. Examining biomarkers in distinct military subgroups may also be important to better delineate clinical utility in combat and combat training settings.

Conclusions
This study found that blood levels of biomarkers for mild TBI and concussion are high after concussions incurred during combative training among MSA cadets. In particular, the GFAP and UCH-L1 biomarkers indicate acute increases with UCH-L1 levels decreasing by 24 to 48 hours after injury and GFAP increases persisting for several days. These biomarkers also had different associations with clinical measures of recovery. Neurofilament light chain levels were generally higher in the concussion group compared with the contact-control group, suggesting that the NF-L biomarker may be measured more remotely after a concussion, although its temporal pattern in the concussion group compared with the contact-control group warrants further investigation. Bloodbased TBI biomarkers have substantial potential to inform our understanding of pathophysiological changes after mild TBI and concussion among military personnel. Future associations with long-term outcomes, impact data, and advanced neuroimaging findings are all opportunities to better delineate the roles of these biomarkers.