Introduction

Mild traumatic brain injury (mTBI) is a stressful condition that activates the hypothalamic-pituitary-adrenal axis (HPA axis), resulting in the release of cortisol1,2. Although acute elevations of this hormone may exert protective immunosuppressive effects, chronically increased levels, either pre or post-mTBI are considered detrimental to recovery processes by exacerbating excitotoxicity, neuroinflammation, and production of reactive oxygen and nitrogen species (i.e., oxidative-nitrosative stress)2. Cortisol has widespread effects on brain regions involved in cognition and emotion, which might influence the persistence of post-traumatic symptoms (PTS) after mTBI3,4,5. In a recent study, blunted salivary cortisol release patterns upon awakening were associated with more severe symptoms after mTBI in college students, possibly via disruptions of the circadian biological clock6. This indicates that the involvement of cortisol in the pathophysiology of mTBI is complex and warrants further, detailed research.

Accumulation of cortisol in hair allows for a retrospective analysis of HPA axis functioning over a longer period of time, including the pre-injury period in case of a mTBI7. In an exploratory study of 46 patients with mTBI performed by our research group, no changes in cortisol levels from pre- to post-injury were found, with similar cortisol levels in a control group8. However, lower pre- and post-injury cortisol levels were associated with higher use of passive coping8, which is considered maladaptive in recovering from mTBI9. Personality traits further determine someone’s ability to cope with stressful situations, and neuroticism in particular is associated with maladaptive coping and developing PTS10,11,12. Moreover, neuroticism, beyond other personality traits, has been linked to higher hair cortisol levels13.

The primary aim of this study was to analyze hair cortisol from pre- to post-injury. We re-explored the hypothesis that hair cortisol levels increase from pre- to post-injury, reflecting an increased stress response to the injury, which would not be present in healthy controls (HC). Additionally, we investigated the relationship of hair cortisol with persistent PTS and functional recovery at six months post-injury, and whether there is an interaction with neuroticism. A final aim was to assess whether mTBI patients who are at higher risk of developing persistent PTS, as defined by a modified version of the Post-Concussion Symptoms Rule (PoCS Rule)14, display different levels of cortisol than patients in the low-risk group.

Methods

Participant inclusion

Participants represent a subcohort of the mTBI patients enrolled in the AIM-TBI study (Dutch trial register number NL8484), a prospective longitudinal multicenter cohort study performed between January 2020 and January 2022. Patients with mTBI (age > 18 years) were included upon presentation at the emergency department (ED) of a level 1 trauma center. A diagnosis of mTBI was made according to the 1993 definition of the American Congress of Rehabilitation Medicine15. Exclusion criteria were major neurologic or psychiatric comorbidity, admission for prior TBI, drug or alcohol abuse, cognitive problems, language barriers and illiteracy. Acute head computed tomography (CT) scans were performed and classified according to the Marshall criteria16. A matched cohort of HC based on age, sex and education, without prior history of TBI were enrolled using the same eligibility criteria as for the mTBI group.

The AIM-TBI study was approved by the Medical Ethics Review Committee of the UMCG (METc 2018/681). Written informed consent was provided by all participants included in this study. All procedures were carried out in accordance with the relevant guidelines and regulations as well as the declaration of Helsinki.

Hair sample collection and analysis

Given the rate of hair growth of approximately 1 cm per month17,18,19, hair samples of at least 3 cm were collected 4–6 weeks after injury. Hair was obtained from the posterior vertex as close to the scalp as possible. Based on the estimated location of the injury along the hair strand, a distal segment of 1 cm (pre-injury) and a proximal segment of 1 cm (post-injury) were determined, each corresponding to the pre- and post-injury moment respectively. Hair samples were further processed as described previously20. In short, samples were weighted and washed, and then grinded using a ball mill (the latter was not done in our previous study8). After methanol extraction, samples were analyzed using tandem liquid chromatography-mass spectrometry (LC-MS/MS; lower limit of quantification = 0.35 pg/mg). Using questionnaires, data were acquired regarding potential confounders for cortisol such as the frequency of hair washing, recent use of corticosteroid medication, presence of dye/bleach/perm products in the hair three months prior to the head injury.

Clinical follow-up measures

Severity of PTS (sum of pre- to post-injury change scores) were measured at two weeks and six months post-injury using the Head Injury Symptom Checklist (HISC)21. The Hospital Anxiety and Depression Scale (HADS) was used to investigate anxiety and depression at two and six months post-injury22. Personality traits were measured at two weeks post-injury using the Neo Five Factor Inventory questionnaire (NEO-FFI)23, and neuroticism was selected as personality trait of interest. Functional recovery was measured at six months post-injury, using the Glasgow Outcome Scale-Extended (GOS-E)24. For statistical analyses, GOS-E scores were dichotomized into complete (GOS-E = 8) or incomplete recovery (GOS-E < 8).

Statistical analysis

Statistical analyses were performed using Python (v 3.10) and associated libraries. Demographics were analyzed using independent sample t-tests or Mann-Whitney U tests, depending on normality, and with Chi-square tests.

Z-scored cortisol concentrations above 3 or below − 3 were considered outliers and removed from further analyses. After visual inspection of distributions, a log10 transformation was applied to the cortisol concentrations to approximate a normal distribution. A linear mixed model (Group [mTBI vs. HC] × Time [pre- vs. post-injury]), with random subject intercepts, was used to analyze the effect of mTBI on cortisol levels. Residuals were inspected for normality. Results were considered significant at p < 0.05. Cohen’s d effect sizes were computed. An additional model was run to account for potential effects of age, sex, use of corticosteroid medication, hair products, and frequency of hair washing.

The relationships of pre- and post-injury cortisol levels with outcome at six months post-injury (dependent variables: PTS and functional recovery), as well as the interaction with neuroticism scores, were investigated with multivariate regression analyses (linear and binomial, respectively). Before implementation, patients with incomplete data were removed, resulting in n = 50 patients included in the model for persistent PTS and n = 56 patients in the dichotomized functional recovery status model. These results were considered significant at a Bonferroni corrected p < 0.05/2 = 0.025.

Lastly, based on the study by Le Sage et al.14, mTBI patients were divided into two risk groups (high/low risk of developing persistent PTS) based on the following characteristics: age, sex, history of TBI and mental health disorder, headache and neck pain upon presentation at the ED and CT abnormalities (Fig. S1). We made no distinction with regards to prior TBI within less than a year (vs. all prior TBI), and neck pain (as measured using the HISC at 2 weeks) was included instead of cervical sprain. Only the ED criteria were used, without the 7-day follow-up variables. Welch’s t tests for unequal samples were used to compare cortisol concentrations between the high- and low-risk groups.

Figures were made using R’s ggplot2 package25,26.

Results

Participants

Hair samples of 66 mTBI patients and 25 age, sex and education matched HC were collected (Table 1). For four patients with mTBI, pre- or post-injury cortisol levels were unavailable due to insufficient hair length or an inadequate number of hairs per segment. Two cortisol outliers (one mTBI and one HC) were omitted (both segments). A total of 61 mTBI patients and 24 HC were included in the final analyses. Groups were similar regarding age, sex and education level, which was coded with a value of 1 to 7, ranging from no completion of elementary school (1) to university graduate (7)27. Analysis of clinical measures (PTS, emotional distress, and functional recovery) for patients with mTBI can be found in the Supplementary Materials.

Table 1 Demographical and clinical characteristics of the mTBI and HC groups.
Full size table

Cortisol measures

In Fig. 1, cortisol concentrations for the mTBI and HC group are depicted (also see Table S1 for central tendency values for raw and log transformed cortisol). A linear mixed model showed a significant main effect of time (p = 0.01, β[SE] = 0.13[0.05], 95%-C.I.: 0.03, 0.24), which was reflected by an increase in cortisol from pre- to post-injury (d = 0.27 and 0.26 for mTBI and HC, respectively). There was no significant main effect of group (p = 0.87, β[SE] = 0.02[0.09], 95%-C.I.: -0.15, 0.18; d = 0.29 and 0.2, for pre- and post-injury cortisol, respectively) and no significant group × time interaction (p = 0.34, β[SE]=-0.06[0.06], 95%-C.I.: -0.19, 0.06).

Fig. 1
figure 1

Scattered boxplots showing log10 transformed cortisol concentrations for healthy controls (HC) and patients with mild traumatic brain injury (mTBI).

Full size image

Results did not change (time: p = 0.01, β[SE] = 0.13[0.05], 95%-C.I.: 0.03, 0.24; group: p = 0.92, β[SE]=-0.01[0.09], 95%-C.I.: -0.19, 0.17); and group × time interaction: p = 0.32, β[SE]=-0.06[0.06], 95%-C.I.: -0.19, 0.06) when including age, sex, hair washing frequency, hair treatment (dye/bleach/perm), and use of corticosteroid medication as nuisance factors/covariates.

Furthermore, there were no differences in pre- (t = 0.44, p = 0.67) or post-injury (t = 1.43, p = 0.16) cortisol levels between patients with and without lesions on CT.

Prediction models for persistent PTS and functional recovery

Pre- and post-injury cortisol levels were not predictive of long-term PTS or functional recovery, and there were no significant interactions with neuroticism scores. Additionally, absolute change (delta [pre vs. post]) of cortisol concentrations were not related to outcome measures. Model parameters are listed in Table S2.

Risk stratification model according to the PoCS rule

Of the 61 mTBI patients, 24 were part of the high persistent PTS-risk group and 37 patients were part of the low-risk group. There were no significant differences in cortisol levels between the two risk groups.

Discussion

Neither at pre- nor at post-injury, there was a significant difference in hair cortisol between patients and HC, which may indicate that hair cortisol is not an adequate measure of the amount of psychological and physiological stress patients with mTBI experience. Although the pre- to post-injury rise in cortisol levels is in accordance with the initial hypothesis, no changes were expected for the controls. In our previous study8, there was a similar effect in both groups, which did not reach the conventional level of significance (0.05 < P < 0.1). The current study contained a greater sample size, and, unlike our previous study, an additional preprocessing step was included (ball mill method) prior to methanol extraction, which can result in higher cortisol concentrations28,29. A potential explanation for higher post-injury vs. pre-injury cortisol levels in both groups is the occurrence of a “wash-out” effect30,31. Multiple factors can contribute to wash-out effects, such as frequency of hair washing, exposure to sunlight and artificial ultraviolet radiation30,32,33,34. In our study, there was no significant main effect of washing frequency in the additional model, and the main effect of time remained significant (i.e., still containing unique variance), suggesting that a possible wash-out effect may have impacted our findings via cumulative effects of multiple (unmeasured) factors.

Hair cortisol concentrations were not significantly associated with long-term PTS or functional recovery status, and there was no interaction with neuroticism. Also, patients at risk for developing persistent PTS did not have higher cortisol concentrations relative to lower-risk patients based on a modified PoCS rule14. Hair cortisol, as measured in our study, represents only a narrow window centered around one acute, complex physiological stressor (i.e., mTBI), followed by possible psychological changes in the later phases post-injury. A short period of exposure to this stressor, combined with a gradual cessation of acute mental stressors, may not be a long enough time-period of above-threshold cortisol levels for it to be detected in hair samples. Cumulative effects of more chronic stress associated with life events prior to mTBI exposure may also superimpose onto this window, making it even more difficult to distinguish the effects of mTBI, if at all present, from other forms of stress. There is increasing evidence that fluctuations in hair cortisol levels around the (individual) baseline can occur across longer periods of time at a slower pace going beyond the effects of degradation and seasonal changes35. This might indicate that longer segments of hair are needed to examine the relationship between cortisol and outcome post-mTBI, although washout effects then become an even bigger issue. This is further supported by the fact that personality characteristics tend to be relatively stable over time36,37, also indicating that this is a chronic measure of stress-regulation capacity. Finally, developmental changes, such as transition from puberty into adulthood, are marked by different patterns of HPA axis activity and associated cortisol levels, possibly influenced by psychological factors, emphasizing the increased complexity when interpreting the role of cortisol in younger mTBI populations38.

Limitations

Our study includes several (additional) limitations which need to be mentioned. Given the relatively small sample size and lack of a standardized reference range for cortisol, it remains difficult to accurately determine whether more extreme cortisol values are true outliers or not. Additionally, metabolic factors such as BMI-index or hip-waist ratio, and solar radiation have not been accounted for in the present study39,40. Furthermore, although the generally accepted hair growth rate is ~1 cm/month, inter-individual differences exist17,18,19, which were not accounted for in the current study.

Conclusion

Altogether, our findings cast doubt upon the applicability of hair cortisol in mTBI research and clinical practice. This is the second study to report null findings on the effect of mTBI on hair cortisol concentrations. However, we acknowledge that the current study may (still) have been limited by low statistical power, imprecision in identifying confounding factors, and uncertainty in establishing reference (control) values. We recommend future investigations focus on more severely injured samples, and the influence of confounding factors.