Higher Cerebrospinal Fluid Levels of Amyloid-β40 Following Traumatic Brain Injury Relate to Confrontation Naming Performance

INTRODUCTION

Nearly 6.2 million Americans are currently living with Alzheimer’s disease (AD), a figure projected to double by 2060 [1]. This condition does not currently have an available cure, and symptoms become increasingly challenging to manage in the advanced stages of the disease. However, AD biological processes start to unfold decades before symptom onset and clinical manifestation [2, 3], underscoring the importance of determining risk factors well before a formal diagnosis is made. Environmental factors are thought to play a pivotal role in enhancing the susceptibility to late-onset AD [4], and traumatic brain injury (TBI) is one particularly important factor associated with increased AD risk [5]. Given our rapidly aging population, it becomes increasingly important to understand the role that TBI plays in conferring risk for AD pathology. Nevertheless, the neurobiological mechanisms underlying this relationship remain unclear [5].

Postmortem work has shown that AD is associated with the accumulation of neuropathological amyloid-β (Aβ) plaques [6], triggering a cascade of subsequent brain changes and eventually leading to cognitive decline and symptoms of dementia [7]. The amyloid hypothesis suggests that this catastrophic plaque aggregation may be caused by early Aβ overproduction via dysregulation within the amyloid-β protein precursor (AβPP) pathway, in which AβPP is cleaved to form Aβ peptides [7, 8]. Importantly, both human and animal studies have shown that increased cerebrospinal fluid (CSF) levels of amyloid peptides (Aβ₄₂, Aβ₄₀, Aβ₃₈) may be indicative of this initial Aβ dysregulation years before Aβ plaques characteristic of AD begin to aggregate [2, 9]. Shorter amyloid species (i.e., Aβ₄₀, Aβ₃₈) tend to be more soluble and thus more abundant in CSF yet less prone to plaque aggregation than Aβ₄₂ [10, 11]. At later stages, in vivo postmortem measurements show that CSF Aβ tends to decrease as Aβ sequestration onto plaques increases [12]. Because the hydrophobic and fibrillogenic Aβ₄₂ peptide is the principal constituent of neuritic plaques [13, 14], symptomatic AD patients typically show decreases in CSF Aβ₄₂ [15]. Taken together, these findings suggest that shorter peptides (i.e., Aβ₄₀, Aβ₃₈) may be pertinent in pre-symptomatic phases of AD prior to plaque aggregation, whereas Aβ₄₂ may be more important in later phases of Aβ dysregulation when plaques develop and AD symptoms appear. Consistent with this timeline and biochemistry, machine learning models have found that increases in soluble CSF Aβ₄₀ and Aβ₃₈ peptides, but not Aβ₄₂, predict downstream neurodegenerative markers in cognitively unimpaired adults, suggesting that early CSF increases in these shorter peptides may be important [16].

Early Aβ overproduction may be a secondary response to damage, such as TBI [8]. Indeed, human autopsy and rodent studies have shown histopathological evidence of altered Aβ₄₀ and Aβ₄₂ levels [17] and AβPP overexpression [18, 19] following TBI of varying severity. However, extant evidence regarding the impact of TBI on CSF Aβ levels is conflicting, with studies showing evidence of both decreased [20, 21] and increased CSF Aβ₄₀ and Aβ₄₂ following TBI [22, 23]. Importantly, most existing studies focused on severe TBI in the acute phase post-injury [21–23] or only included young adults [20], thus hindering interpretations about AD-related biomarker changes in older adults following a remote history of TBI of varying severity.

AD-related clinical signatures are often captured using cognitive measures sensitive to early AD impairments, including verbal memory and semantic processing [24, 25]. Although verbal memory deficits have also been linked to mild TBI, semantic abilities are relatively resistant to TBI [26, 27]. Therefore, the juxtaposition of these two cognitive measures affords the ability to test whether pathological changes following TBI may implicate cognitive change that is either mainly or jointly related to AD.

The current study assessed whether CSF levels of Aβ₄₂, Aβ₄₀, and Aβ₃₈ were elevated in non-demented Veterans with a history of TBI of any severity compared to those without a TBI history. Confrontation naming and memory performance were compared between groups, as measured by the Boston Naming Test (BNT) and Rey Auditory Verbal Learning Test (AVLT), to highlight any early preclinical cognitive change. Finally, we investigated the influence of CSF Aβ on the relationship between TBI history and cognition through mediation models. We hypothesized that the TBI group would have higher CSF Aβ levels in addition to worse cognitive scores versus those without a TBI history. We also expected CSF levels of Aβ to mediate the relationship between TBI history and cognition.

MATERIALS AND METHODS

Participants

Figure 1 outlines the inclusion criteria used to generate our sample of 112 Veterans from the ADNI-DOD data repository. To investigate the effects of TBI on cognition and fluid-based Aβ levels, participants were selected based on completion of a self-report questionnaire assessing lifetime TBI history and cognitive testing, as well as availability of CSF Aβ biomarker data. Exclusion criteria for this study included penetrating head injuries, loss of consciousness due to drug overdose, and unknown or contradictory TBI reports. ADNI-DOD also excluded participants with clinically diagnosed mild cognitive impairment/dementia to minimize the additive effects of TBI on existing pathology. Additional details regarding inclusion/exclusion criteria specific to ADNI-DOD can be found in the protocol (https://adni.loni.usc.edu/wp-content/uploads/2013/09/DOD-ADNI-IRB-Approved-Final-protocol-08072012.pdf).

OPEN IN VIEWER

Fig. 1

Flowchart of inclusion/exclusion criteria used to select sample. TBI, traumatic brain injury; CTRL, controls; MSMS, tandem mass spectrometry; CSF, cerebrospinal fluid; Aβ, amyloid-beta; BNT, Boston Naming Test.

The final sample of Veterans was categorized into two groups: those with a lifetime history of TBI (TBI = 56) and controls (CTRL = 56). Lifetime TBI history was assessed using The Ohio State University TBI Identification Method – Interview Form, a psychometrically validated structured interview that is currently the gold standard for retrospectively assessing lifetime TBI history [28, 29]. We included individuals with a self-reported history of head/neck injuries that occurred before, during, and after the Vietnam War to capture the full range of TBI subtypes, rather than strictly combat- or blast-related TBI. TBI severity was classified based on the VA/DOD criteria [30]. Mild TBI was defined as loss of consciousness < 30 min, posttraumatic amnesia≤24 h, and/or altered mental state≤24 h. Moderate/severe TBI was defined as loss of consciousness≥30 min, posttraumatic amnesia > 24 h, and/or altered mental state > 24 h. These moderate and severe ratings were pooled because questionnaire items did not allow proper discrimination between the two severity levels. If more than one severity category was met in cases with multiple TBIs, the higher severity (i.e., moderate/severe TBI) was assigned.

RESULTS

Demographic characteristics and neuropsychological scores of TBI and CTRL groups are described in Table 1. The two groups did not significantly differ in sex or race distributions, education, age, or the CSF-to-Test interval. Although the TBI group performed slightly worse than the CTRL group on BNT and AVLT, these differences were not significant (all ps > 0.10).

Within the TBI group, injury severity was relatively balanced (n = 25 mild, n = 28 moderate/severe, n = 3 missing data required to calculate injury severity). Because severity data was unavailable for three participants, we combined all severities in subsequent analyses to increase power. Injury frequency, defined as the total number of lifetime TBIs reported by an individual, ranged from one to six, with a majority experiencing only a singular TBI (n = 30 with one TBI, n = 14 with two TBIs, n = 9 with three or more TBIs, n = 3 with inconclusive injury frequency data). The median timeframe between participants’ latest TBI and CSF collection date (TBI-to-CSF) was 44 (IQR: 37–47) years.

Across the entire sample, Aβ peptides were highly correlated with one another. CSF Aβ₄₂ was positively associated with both CSF Aβ₄₀ (r = 0.74, p < 0.0001) and CSF Aβ₃₈ (r = 0.71, p < 0.0001). CSF Aβ₄₀ and CSF Aβ₃₈ levels were also strongly related (r = 0.96, p < 0.0001).

Table 1 also summarizes comparisons of CSF Aβ concentrations between groups. The TBI group had significantly higher levels of CSF Aβ₃₈ (t = –2.10, p = 0.038) (Fig. 2A) and CSF Aβ₄₀ (t = –2.43, p = 0.017) (Fig. 2B) but only marginally greater levels of CSF Aβ₄₂ (t = –1.77, p = 0.079) than the CTRL group (Fig. 2C). There was no significant groupwise difference in Aβ₄₂/Aβ₄₀ ratios (Fig. 2D). Within the TBI group, CSF Aβ concentrations did not differ based on injury frequency (all ps > 0.1).

OPEN IN VIEWER

Fig. 2

CSF Aβ levels between TBI and CTRL groups. Participants with a history of TBI have higher levels of CSF Aβ₃₈ (A) and Aβ₄₀ (B) than CTRL participants. There was no significant group difference in CSF Aβ₄₂ (C) or Aβ₄₂/Aβ₄₀ ratio (D). Violin plots illustrate the distribution of data and provide the median, interquartile range, and range of CSF concentrations for each group independently.

Because CSF Aβ₄₀ and Aβ₃₈ concentrations were significantly different between groups, we correlated these CSF measures with neuropsychological performance. We found negative relationships between BNT and both Aβ peptides (Aβ₄₀: rho = –0.20, p = 0.032; Aβ₃₈: rho = –0.19, p = 0.048) (Fig. 3). However, we found no relationship between Aβ peptides and AVLT-Imm or AVLT-Del(all ps > 0.1).

OPEN IN VIEWER

Fig. 3

CSF Aβ levels correlate with BNT performance. Scatterplots depict individual participants’ Boston Naming Test (BNT) scores plotted against their concentrations of CSF Aβ₃₈ (A) and Aβ₄₀ (B) with a line of best fit.

We interrogated these significant relationships further by testing the mediating effects of Aβ₄₀ and Aβ₃₈ on the relationship between self-reported TBI history and BNT performance. The first model revealed a significant indirect effect of TBI on BNT performance through Aβ₄₀ (β= –0.16, Bootstrapped 95% CI [–0.393, –0.004]), which accounted for roughly half of the total effect (P_M = 0.54). Sustaining a TBI was positively associated with Aβ₄₀ levels (β= 0.43, p = 0.022), and in turn, Aβ₄₀ levels negatively affected BNT performance (β= –0.14, p = 0.046). The direct effect of TBI on BNT was not significant (p > 0.1) (Fig. 4). Direct and indirect effects of TBI on BNT through Aβ₃₈ were not significant (both ps > 0.1).

OPEN IN VIEWER

Fig. 4

CSF Aβ₄₀ levels mediate the relationship between TBI and BNT performance. TBI history has a significant indirect effect on Boston Naming Test (BNT) performance through CSF Aβ₄₀, such that TBI history is associated with increased CSF Aβ₄₀ levels, which are related to worse BNT performance. ^*Indicates significance at p < 0.05 threshold.

DISCUSSION

This study examined the link between self-reported TBI history, CSF levels of Aβ peptides (Aβ₄₂, Aβ₄₀, and Aβ₃₈), and cognitive performance in Vietnam war Veterans. Three key findings emerged. First, Veterans who reported a history of TBI exhibited higher levels of CSF Aβ₄₀ and Aβ₃₈ compared to those without a TBI history. Second, an increase in CSF Aβ peptide levels correlated with poorer confrontation naming ability, a measure of language skills, but no significant relationship was found with memory performance. Finally, CSF Aβ₄₀ had an indirect effect on the relationship between TBI history and confrontation naming performance. These findings suggest that TBI is associated with early increases in CSF Aβ levels and concomitantly lower language performance at early stages.

Prior research in both human and animal models has established alterations in Aβ levels after TBI [20–23, 42]. Rodent studies showed a rise in tissue levels of Aβ₄₀ and Aβ₄₂ eight weeks after TBI compared to sham CTRLs [41], and a cerebral microdialysis study showed similar trends of increased interstitial Aβ₄₀ and Aβ₄₂ in humans with diffuse brain injury [42]. Nonetheless, previous work examining human CSF Aβ levels following TBI is mixed, with some studies observing reductions [20, 21] and others elevations in CSF Aβ₄₀ and Aβ₄₂ throughout the first week after severe TBI, compared to CTRLs [22, 23]. While these existing CSF studies focused on severe TBI and its immediate repercussions on Aβ₄₀ and Aβ₄₂ dysregulation, the present study shows evidence of prolonged, cross-sectional increases in CSF Aβ₄₀ following a remote history of TBI, across a sample with evenly distributed severity levels of TBI.

Discrepancies in studies showing increased versus decreased CSF Aβ levels following TBI may be due to limitations in distinguishing diffuse from neuritic plaques. AD plaques are predominantly neuritic and develop slowly across many years, whereas diffuse TBI-related plaques can appear rapidly within hours [18, 43]. Nonetheless, human autopsy studies have revealed increased Aβ₄₂ plaque deposits in the brains of severe TBI patients relative to CTRLs [18, 43], and these deposits are similar to those seen in AD [44]. A recent study also showed greater neuritic Aβ plaque burden in the brains of long-term survivors of a single TBI compared to healthy adults, suggesting that acute diffuse Aβ pathology following TBI can remain in the brain and potentially progress to denser plaques characteristic of AD over the course of several years [45]. Taken together, these findings suggest that TBI of varying severity has the potential to induce or accelerate the pathological AD process, as longer-term Aβ dysregulation could ultimately amass to Aβ plaque aggregation characteristic of AD [9–14].

Preclinical AD has previously been associated with early impairments in both confrontation naming and verbal memory [24, 25]. The results reported here suggest that Aβ₄₀ levels may serve as a mechanism by which TBI implicates AD-like changes in confrontation naming abilities, although we found no association between CSF Aβ levels and verbal memory scores. There are two potential explanations for this discrepant finding. First, verbal memory measures may be confounded by age-related cognitive change given our older, non-demented adult sample. Indeed, studies have shown that verbal memory abilities tend to diminish as early as the sixth decade of life [46], whereas confrontation naming abilities tend to be more stable throughout the aging process with only subtle difficulties appearing in the seventh to eighth decades of life [47]. In the current study, groupwise differences in AVLT may have been overshadowed by the effects of aging in both groups, whereas differences in BNT scores likely persisted across groups due to the decreased sensitivity of confrontation naming to effects of aging. Secondly, while both memory and language impairments have been linked to early AD [24, 25], research on the cognitive sequelae of TBI has shown acute and long-term effects on verbal memory impairments yet minimal impacts on language abilities [26, 27]. This dichotomy suggests that the mechanisms subserving TBI-related cognitive change may be independent from those implicating early AD-like cognitive change. Thus, our results, which show significant impacts of language but not memory, imply that early increases in CSF Aβ₄₀ may be capturing a unique pathological process that is not solely driven by TBI-related mechanical insults. Rather, our findings suggest that early increases in CSF Aβ₄₀ may represent a pathological AD process that is triggered by TBI yet independent of TBI-induced cognitive change.

Indeed, Aβ overproduction has been previously established within the context of AD. Early-onset familial AD is caused by specific genetic mutations (e.g., APP, presenilin) involving overproduction of Aβ species [48]. Moreover, individuals with Down syndrome have an extra chromosome 21, which contains the APP gene locus, and a significantly increased risk of AD [49]. Multiple studies have suggested that this heightened AD risk may be due to lifelong overexpression of AβPP, leading to increased production of all three Aβ species (Aβ₃₈, Aβ_40, Aβ₄₂) and subsequent plaque development early in life [50, 51]. Paralleling this comparable overproduction of Aβ across peptides seen in Down syndrome, the present study found similar increases in all three Aβ peptides within our TBI group as well as significant correlations between all three Aβ peptides across our entire non-demented sample. These results corroborate similar linear relationships found among Aβ peptides in neuropathologically-negative individuals defined by CSF tau cut-off indexes [52]. However, this prior work revealed that only CSF Aβ₃₈ and Aβ₄₀ levels, but not Aβ₄₂ levels, remained correlated in neuropathologically-positive individuals with increased tau burden. Taken together, this work suggests that all three soluble Aβ species may share similar coregulation mechanisms yet demonstrate differential fates depending on pathology [52]. Future studies should incorporate other markers of neurodegeneration (e.g., NfL, tau) to better capture pathological disease states and investigate potential connections to Aβ overproduction, especially within the context of TBI.

Despite this extant evidence of global Aβ overproduction and traditional associations of CSF Aβ₄₂ with AD [15], our findings underscore the consideration of CSF Aβ₄₀ in early AD states, especially following TBI. Indeed, recent clinical studies found significant increases in CSF Aβ₄₀ in AD patients relative to controls [53] and other neurodegenerative disease groups [54, 55], and mean levels were comparable to average CSF Aβ₄₀ levels found in the present study, especially in sub-samples using similar biomarker processing methodology (i.e., MS/MS) [55]. This work corroborates the link between increased CSF Aβ₄₀ and AD pathology. However, the present study found no association of TBI with Aβ₄₂/Aβ₄₀ ratios despite apparent influences of TBI on individual Aβ peptides, specifically CSF Aβ_40 .  Because Aβ₄₂/Aβ₄₀ is classically linked with pathological Aβ plaque burden [33, 34], our collective findings could suggest an early, pre-pathological state in which Aβ production via the AβPP pathway is dysregulated yet Aβ pathology has not yet developed [7, 8]. Because Aβ₄₀ tends to be more ubiquitous and less affine than Aβ₄₂ [10, 11], CSF measures may be more sensitive to Aβ₄₀ overproduction in early stages. At later stages, the Aβ₄₂ peptide—with its high aggregational propensity—begins to sequester onto plaques, thus leading to progressive decline characteristic of AD [12–15]. Indeed, extant research in cognitively normal adults has shown evidence of increased baseline CSF Aβ₄₀ levels predicting longitudinal 10-year decreases in CSF Aβ₄₂ and subsequent cognitive decline, further substantiating a potential pre-pathological stage [3]. However, present methodology precludes direct examination of neuropathology, so future studies should incorporate positron emission tomography (PET) imaging to detect in vivo Aβ accumulation to better elucidate AD neuropathological staging.

There are a few limitations to note in this study. First, due to the goals of ADNI-DOD and the cross-sectional nature of the study, we cannot determine clinical progressions to AD or other dementias nor longitudinal changes in CSF Aβ over time. We also cannot identify baseline Aβ levels prior to TBI or immediately following TBI. Future work should examine relationships between TBI, cognitive performance, and CSF Aβ levels longitudinally. Similarly, retrospective lifetime TBI history was collected via self-report methods, which may be prone to recall bias over time. Although a structured clinical interview measure, which represents the current gold standard in the field [29], was used to minimize this bias, subsequent TBI assessment should incorporate medical records and/or collateral information to improve precision of the TBI data. Finally, our sample was relatively small and composed of majority White, non-Hispanic/Latino, male Veterans, leading to small effects and unclear generalization to the broader population. Future efforts should focus on recruiting a larger, more diversified sample to improve effect sizes and external validity.

Overall, these findings suggest that TBI may confer risk for AD through dysregulation of the AβPP pathway and specifically, through its association with increased production of Aβ. Although decreases in CSF Aβ are typically observed in AD [12], the present study underscores the importance of temporal staging in Aβ biomarker assessment and interpretation. TBI might be linked to a surge in CSF Aβ, signaling an overabundance of Aβ in early, non-pathological stages, which could in turn elevate the risk of subsequent AD pathology. Given these findings, there is a clear necessity to devise in vivo biomarker models that account for multiple Aβ peptides. Such models could aid in identifying AD risk post-TBI and expose cognitive vulnerability early in the pathological process.

AUTHOR CONTRIBUTIONS

Erica Howard (Conceptualization; Formal analysis; Investigation; Methodology; Visualization; Writing – original draft); Jena N. Moody (Formal analysis; Investigation; Validation; Writing – review & editing); Sarah Prieto (Conceptualization; Methodology; Writing – review & editing); Jasmeet P. Hayes (Conceptualization; Funding acquisition; Methodology; Resources; Supervision; Writing – review & editing).