Literature Review

Can biomarkers detect concussions? It’s complicated

View on the News

Comprehensive study shows biomarker limitations

Concussion diagnosis has been constrained by reliance on subjective evidence, particularly in mild cases. Concussions also often result from a wide range of injuries, but focusing on sports-related concussions offers a chance to study biomarkers in a more controlled way.

These three studies represent the most comprehensive sports-related concussion biomarker work to date. The message may be that, for sports-related concussions, serum biomarkers may be able to detect the occurrence of a concussion, but they cannot predict motor, neurobehavioral, or neurocognitive outcome measures.

The study results also underline the need for larger, more complex prospective studies.

Erin Bigler, PhD, is a professor of psychology and neuroscience at Brigham Young University. Ellen Deibert, MD, is a neurologist in York, Pa. These comments were taken from an accompanying editorial (Neurology. 2018. doi: 10.1212/WNL.0000000000006609 ). Dr. Bigler and Dr. Deibert have no relevant conflicts of interest.



A series of three studies in college students showed that some serum markers are associated with concussion but the background level of the markers can vary considerably. There was no association between the markers and history of concussion, and they markers varied significantly by sex and race.

jpbcpa/Getty Images

The work, published in Neurology, suggests that there is hope for finding biomarkers for concussion, but much more work needs to be done.

Serum levels of amyloid beta 42 (Abeta42), total tau, and S100 calcium binding protein B (S100B) were associated with concussion, especially when tests were performed within 4 hours of the injury. However, the varying background levels indicate that these biomarkers are not yet ready for clinical application.

All three studies looked at serum levels of Abeta42, total tau, S100B, ubiquitin carboxyl-terminal hydrolase L1 (UCH-L1), glial fibrillary acidic protein (GFAP), microtubule-associated protein 2 (MAP2), and 2’,3’-cyclic-nucleotide 3’-phosphodiesterase (CNPase).

In the first study, researchers recruited 415 college athletes without a concussion (61% male, 40% white). The researchers took measurements outside of the athletes’ competitive sports season to maximize the odds that the levels would represent a true baseline. The median time between blood draw and the last risk of head impact was 80 days (mean, 98.4 days; interquartile range, 38-204 days).

Males had higher levels of UCH-L1 (Cohen d = 0.75; P less than .001) and S100B (Cohen d = 0.56; P less than .001), while females had higher levels of CNPase (Cohen d = 0.43; P less than .001). White subjects had higher levels of Abeta42 (Cohen d = .28; P = .005) and CNPase (Cohen d = 0.46; P less than .001). Black subjects had higher levels of UCH-L1 (Cohen d = 0.61; P less than .001) and S100 B (Cohen d = 1.1; P less than .001).

The measurements were not particularly reliable, with retests over 6- to 12-month periods yielding varying results such that none of the test/retest cutoff points reached the cutoff for acceptable reliability.

The second study was an observational cohort study of the same 415 subjects. The researchers assessed the self-reported concussion history and the cumulative exposure to collision sports with serum levels of the above biomarkers. The only relationship between a biomarker history and self-reported concussions was higher baseline Abeta42, but that had a small effect size (P = .005). Among football players, there was no association between approximate number of head impacts and any baseline biomarker.

The third study looked at 31 subjects who had experienced a sports-related concussion, 29 of whom had had both a baseline and a postconcussion blood draw, and compared them with nonconcussed, demographically matched athletes.

Of all the biomarkers studied, only levels of S100B rose following a concussion, with 67% of concussed subjects experiencing such a change (P = .003). When the researchers restricted the analysis to subjects who had a blood draw within 4 hours of the concussion, 88% of the tests showed an increase (P = .001). UCH-L1 also rose in 86% of subjects, but this change was not significant after adjustment for multiple comparisons (P greater than .007).

Compared with controls, concussed individuals had significantly higher levels of Abeta42, total tau, S100B, and GFAP. Of the concussed patients, 79.4% had Abeta42 levels higher than the median of controls, 67.6% had higher levels of total tau than the median of controls, and 83.3% had higher levels of S100B. Restriction of analysis to blood drawn within 4 hours of the injury yielded values of 81.3%, 75.0%, and 88.2%, respectively.

When limited to blood draws taken within 4 hours of injury, the researchers found fair diagnostic accuracy for measurements of Abeta42 (area under the curve, 0.75; 95% confidence interval, 0.59-0.91), total tau (AUC, 0.74; 95% CI, 0.58-0.90), and S100B (AUC, 0.75; 95% CI, 0.64-0.85). Abeta42 concentrations higher than 13.7 pg/mL were 75.0% sensitive and 82.4% specific to a sports-related concussion. Total tau concentrations higher than 1.7 pg/mL detected sports-related concussions at 75.0% sensitivity and 66.3% specificity, with acceptable diagnostic accuracy for white subjects (AUC, 0.82, 95% CI, 0.72-0.93). Also for white participants, S100B concentrations higher than 53 pg/mL predicted sports-related concussions with 83.3% sensitivity and 74.6% specificity.

The researchers found no associations between biomarkers and performance on clinical tests or time away from sports.

SOURCE: BM Asken et al. Neurology. 2018. doi: 10.1212/WNL.0000000000006613.

Next Article: