Traumatic brain injury (TBI) is a leading cause of disability and death globally.¹ Effective neuroprotective treatments are lacking due, in part, to heterogeneity in demographic and clinical characteristics and variability in acute secondary injury cascades post-injury.^2,3 To inform prognostication and identify neuroreparative treatments, we previously examined inflammatory markers, sex hormones, and monoaminergic and neurotrophic molecules as potential biomarkers indicative of disability and mortality after TBI.^2,4–7 Impaired cognition is one of the most common problems after TBI, occurring in ~70% of individuals with severe injury.⁸ Cognitive deficits broadly affect functioning, disrupting daily life activities, employment, social interaction, and life satisfaction.^8,9 Early biomarkers specific to poor cognitive prognosis may inform outcome prediction and personalized cognitive dysfunction treatment approaches.

Neuroendocrine hormone profiles may be sensitive predictive markers for cognitive deficits after TBI. Acute TBI affects the hypothalamic-pituitary-adrenal (HPA) axis, which regulates the stress response and can mediate negative effects of inflammation on the CNS.^2,6,10,11 In response to a stressor, the hypothalamus releases corticotropin-releasing hormone (CRH) which binds to receptors in the anterior pituitary gland, in turn releasing adrenocorticotropic hormone (ACTH).¹² The adrenal glands then produce cortisol in response to this ACTH signaling. We previously documented elevated cerebrospinal fluid (CSF) and serum cortisol profiles occurring over the first week after severe TBI;^6,10 elevated CSF cortisol was associated with worse 6 and 12-month global and disability outcomes.¹⁰ Elevations can occur from disrupted negative feedback involving the HPA axis and either decreased glucocorticoid receptor sensitivity of HPA axis organs or decreased CRH, both of which may result in sustained elevated cortisol levels.^13–15

Though cortisol can have transient beneficial anti-inflammatory effects, aiding in immune cell removal and inhibiting immune cell infiltration at injury sites in the periphery and CNS,^11,16 elevated cortisol levels can lead to excessive or maladaptive inflammatory responses post-injury¹⁴, perpetuating blood brain barrier (BBB)¹⁷ and neural cell dysfunction.¹¹ We have shown after severe TBI that elevated acute (days 0–5) CSF cortisol mediates the relationship between inflammatory cytokine profiles and 6-month post-injury outcome.² We have also established that CSF BDNF levels and their associated relationships with mortality differ based on CSF cortisol trajectories after TBI.¹⁸ Serum cortisol levels reflect HPA axis response to TBI and associated injuries, since the adrenal glands are the primary contributors to systemic cortisol levels post-injury.^6,10,21 Systemic cortisol transport into the CNS is the primary source of brain exposure to cortisol. This transport is highly regulated, but cortisol can also move passively into the CNS in the setting of BBB compromise.^17,22,23 Our work shows CSF cortisol levels after TBI can be up to 10X that of healthy controls, and higher CSF cortisol levels measured from samples collected during the first week after injury are associated with mortality and survivor-specific functional outcomes after severe TBI.^10,18 However, no one has explored how acute cortisol profiles affect cognition among TBI survivors. Our objective was to identify relationships between days 0–5 post-injury serum and CSF cortisol profiles and 6-month neuropsychological test scores after severe TBI. We hypothesized that consistently high serum and CSF cortisol profiles are biomarkers associated with poor cognition measured six months post-TBI.

The novel association we found between acute cortisol profiles and later cognitive impairment after severe TBI suggests cortisol may be a potential early treatment target to improve long-term cognition post-TBI. Our results indicate those with acute serum and CSF cortisol profiles that are high and remain high over days 0–5 post-injury have poorer cognition at 6-months post-injury than those whose cortisol profiles are consistently low or may start high but decline early post-injury. This finding suggests acute cortisol trajectories among TBI survivors are an early indicator of ongoing hypercortisolemia, which subsequently affects later cognition. If these results are confirmed, early cortisol trajectories could be used to identify “at-risk” individuals who may benefit from therapies that modulate cortisol production and signaling.

In the absence of TBI, HPA axis feedback loops associated with elevated serum cortisol levels signal to the pituitary and hypothalamus to decrease ACTH and CRH production, thus decreasing further cortisol production. However, multiple cytokines (e.g. IL-6, TNFα) can contribute to sustained cortisol elevations³⁶ such as what occurs in response to trauma,^2,13 but cortisol elevations can also result from reduced glucocorticoid-mediated feedback to the HPA axis.¹⁵ This latter phenomenon has been examined in rats, where HPA feedback may be disrupted by changes in glucocorticoid receptor expression.³⁷ Although cortisol has been extensively studied, little evidence exists regarding mechanistic cortisol effects peripherally or centrally after TBI. Cortisol production occurs primarily in the adrenal glands,²¹ although some CNS-derived hormone synthesis may occur.²¹ Further, excess serum cortisol may enter the CNS after TBI due to P-glycoprotein dysfunction, an important BBB transporter that limits accumulation of drugs and hormones like cortisol in the brain.^17,22,23 Our findings align with this research, in that those with higher serum and CSF cortisol profiles had worse cognition.

Relevant to the acute stress response associated with TBI, differential cortisol receptor binding may contribute to how stress-induced elevated cortisol levels affect cognition.³⁸ Cortisol binds with both mineralcorticoid (MR) and glucocorticoid receptors (GR).³⁹ MR’s have a higher affinity for cortisol and are the main receptors maintaining cortisol mediated circadian rhythm.³⁹ Cortisol receptors are widely distributed in the CNS, wherein MR’s are present in the limbic system and GR’s are present in both subcortical and cortical regions, and affect cognition.³⁹ Differential receptor occupation during stress negatively affects hippocampal function. During stress-induced hypercortisolemia, similar to accumulation of cortisol acutely after TBI, MR’s become saturated, resulting in increased GR binding.³⁹ Experimental TBI work in fluid percussion has demonstrated aberrant stress responses after injury,^13,40 yet the work linking dysfunctional stress responses after TBI is both complex and likely temporally and receptor dependent. Previous experimental work suggests GR inhibition with mifepristone can mitigate a disrupted neuroendocrine stress response.⁴¹ Other experimental studies suggest MR stimulation, in the setting of reduced plasma corticosterone, improves spatial memory early after TBI⁴² and that GR down-regulation and MR upregulation are both necessary in the setting of low endogenous corticosterone levels to promote neuronal survival and spatial learning and memory early after TBI.⁴³ Future work is needed to elucidate how cortisol stress responses are disrupted clinically after TBI and how CNS GR/MR balance in TBI recovery affects learning and memory over time. Future work may also require specific focus on age and sex HPA response moderators,^44,45 in appropriately sized TBI and control cohorts, particularly given some indication in our survivor data that age and sex affect early cortisol profiles post-TBI.

Imaging studies characterizing CNS cortisol effects on specific brain structures suggest structure-specific vulnerability to elevated cortisol levels. One study using pharmacological functional magnetic resonance imaging found that acute hydrocortisone administration resulted in a time dependent increase in hippocampus blood-oxygen-level dependent signals, indicating cortisol may have rapid effects on structures like the hippocampus.⁴⁶ Another studied reported that long term effects of excessive cortisol exposure due to Cushing’s syndrome led to brain volume loss and grey and white matter abnormalities that affect the hippocampus and medial frontal and anterior cingulate gyri.⁴⁷ These changes may be reversible after cortisol level correction.^19,47 Given knowledge of ongoing brain volume atrophy in the context of TBI³³ and our findings of acute hypercortisolemia, future work should evaluate if acute cortisol exposure contributes to or accelerates TBI-related brain atrophy. The role of persistent hypocortisolemia on brain atrophy post-TBI should also be explored. Prior reports using a larger cohort from our center suggest that acute adrenal insufficiency (AI) can occur after severe TBI, but AI status did not adversely affect acute care mortality, vasopressor or corticosteroid use, or six month global outcome.⁶

Although our work only evaluates cortisol levels over days 0–5 post-injury, our data suggest high cortisol levels early after TBI may portend ongoing hypercortisolemia that contributes to long-term cognitive impairments. If true, hypercortisolemia may affect entire brain network function after TBI.⁴⁸ More research is needed to examine how cortisol affects cognitive neural networks, and associated vulnerable structures, after TBI. For example, elevated levels may persist and lead to ongoing increases in GR binding involving brain regions relevant to cognition. Differential binding affinities associated with MR’s versus GR’s may leave regions with higher GR densities relative to MR’s vulnerable to damage due to hypercortisolemia.⁴⁹ Further, adverse effects of hypercortisolemia on plasticity and neurogenesis mechanisms after TBI may result from specific receptor activation. For example, MR agonists can increase LTP while GR activation can decrease LTP.⁵⁰ TBI recovery relies on injury-induced increases in plastic activity to support brain repair, which we hypothesize may be hampered in the setting of hypercortisolemia.⁵¹

We note that trajectory groups in both serum and CSF have cognitive composite scores below normal (<50), and we describe domain-specific and total cognitive composite scores at levels considered cognitively impaired (<40). However, the lower scores in the high trajectory groups suggest a cortisol dose-response effect. Clinically meaningful differences in cognitive testing may generally be detected at 0.5 standard deviation differences,⁵² representing a 5 point difference in the current study, which we observed between trajectory groups in several domains. Memory domain performance in our study was impaired in both high and low trajectory groups but were not different between groups in serum or CSF, suggesting there may be a ceiling effect for cortisol with respect to memory relative to other domains. Memory impairment also is fairly nonselective and universal after moderate-to-severe TBI,⁵³ potentially explaining why cortisol may have less specific direct effects on memory compared to other domains. Other reasons for memory being insensitive to acute cortisol profiles may include the neuropsychological tests selected for this study, sample size, and the exclusion of those too severely injured to complete testing.

Higher education and hospital LOS, were also associated with cognitive performance. Higher education has been associated with more cognitive reserve and better outcomes after TBI,⁵⁴ and longer LOS has been associated with more severe injuries and with older age,⁵⁵ both of which are consistent with our findings. Contusion was marginally associated with both CSF and serum cortisol levels and also with cognitive composite scores. Although our review of the literature finds no evidence that contusions increase cortisol levels, one might surmise that the intraparenchymal blood and lesion burden associated with contusion might affect cortisol levels. However, multivariable analyses show that CSF and serum cortisol TRAJ remained significantly associated with cognitive composite scores, even when adjusting for contusion. Thus, despite the postulation that intraparenchymal blood associated with contusions may affect CSF and serum cortisol, our data show that cortisol independently affects cognition.

This study has several limitations and future directions to consider. Our cohort was limited to survivors with acute cortisol levels and 6-month cognitive testing. Individuals cognitively unable to complete neuropsychological testing were excluded, so cognition in this cohort may not accurately represent more severe cognitive dysfunction. Since this study only examines acute cortisol profiles in patients, we cannot conclude if early elevated cortisol directly causes these cognitive deficits or simply reflects the start of a chronic hypercortisolemic state that contributes to cognitive impairment. However, rates of chronic central hypoadrenalism range from 5% to 46% after TBI,⁵⁶ suggesting that hypercortisolemia is likely to be a time-limited phenomenon and that acute cortisol exposure may contribute significantly to cognition, in its own right, after TBI. Yet, it remains unknown how acute cortisol profiles correspond to cortisol measurements in subsequent months post-injury,⁵⁶ particularly in the setting of ongoing inflammation post-TBI.⁵⁷ Unmeasured health and environmental post-acute factors may also influence stress and cortisol levels, including socioeconomic challenges, caregiver support, or other adverse life or health events occurring between initial hospitalization due to TBI and 6-month outcomes. As an observational study, we also cannot determine if cortisol profiles directly influence cognition or are indicators of other processes affecting cognition. We assessed serum/CSF cortisol levels, but hair or urinary cortisol may offer more stable measurements that reflect total cortisol exposure over time, particularly when monitoring TBI recovery. Salivary cortisol profiles post-TBI may be feasible in tracking diurnal cortisol variation for those cognitively able to complete this type of complex testing in home and community settings.

Cortisol secretion is not the only stress response factor that may affect cognition. For example, catecholamines can be affected by stress,⁵⁸ and alterations may perpetuate cognitive decline in the setting of TBI.^7,34 Since inflammatory signaling regulates both HPA and catecholamine pathways, inflammatory pathway relationships to both catecholamine and cortisol production over time should be assessed together for their independent and interrelated impacts on cognitive function after TBI. HPA axis function, inflammation, and catecholaminergic tone can all be modulated by lifestyle interventions like yoga⁵⁹ and mindfulness,⁶⁰ presenting therapeutic non-pharmacologic, rehabilitation-relevant interventions for at-risk individuals that may improve cognitive outcomes after TBI.