Effect of Traumatic Brain Injury on Sensorimotor Gating

Abstract

Background:

Traumatic brain injury (TBI) can impair acute and long-term cognitive functions and increase risk for developing neuropsychiatric disorders, including mood disorders and post-traumatic stress disorder. Sensorimotor gating, operationalized by prepulse inhibition (PPI) of the acoustic startle response, is impaired across several neuropsychiatric disorders and can contribute to neuropsychiatric risk. Although animal models indicate TBI can robustly disrupt PPI, no well-powered studies exist in humans.

Methods:

We examined associations between TBI history and sensorimotor gating in the Marine Resiliency Study, a cohort of active-duty Marines and Navy Corpsmen (N=1725). Participants reported lifetime TBI history, injury severity, blast exposure, time since last injury, number of TBIs, and current TBI-related problems. Individuals completed PPI and acoustic startle habituation assessments. Analyses were adjusted for neuropsychiatric symptoms, hearing, and baseline startle.

Results:

Moderate to severe TBI (N=128), but not mild TBI (N=824), was associated with lower PPI vs controls (N=773). Effects were most prominent at shorter interstimulus intervals (30-60msec), which may reflect more alterations to automatic/preconscious processing periods. TBI-associated disruption of PPI was most strongly associated with reported loss of consciousness >30 minutes. Blast-associated TBI was uniquely associated with reduced startle habituation, but not PPI deficits. Sensorimotor gating effects were independent of self-reported problems after TBI.

Conclusions:

Moderate-severe and blast TBIs are associated with reductions in sensorimotor gating years after injury, supporting PPI as a useful cross-species marker of TBI-associated functional disruption with translational utility. Longitudinal studies are needed to determine contributions of TBI-associated PPI disruptions to cognitive deficits and neuropsychiatric vulnerability.

Introduction

Traumatic brain injuries (TBI) are highly prevalent in the US, with recent data suggesting >18% of the population has sustained at least one TBI in their lifetime and 5.3 million have a disability due to TBI [1]. TBI-associated healthcare utilization costs are significant, with an estimated $40 billion dollars annually for cumulative costs for non-fatal injuries [2]. TBI is a hallmark injury of military conflict and training, with increased exposure to explosive devices, mortars, artillery, and blast-exposure during routine breacher trainings being a cause of major disability [3]. Acute TBI symptoms include post-traumatic amnesia (PTA), altered mental state, loss of consciousness (LOC), and/or intracranial lesions [4,5]. Severity classification is largely based on endorsement of these symptoms, with a LOC<30 minutes considered “mild” [4,5]. Although most TBIs are classified as “mild” [4,6,7], between 20-50% of TBI patients experience long-term symptoms [8–12]. Although acute symptoms following initial injuries can subside over time, TBI increases risk for developing neuropsychiatric disorders, including mood disorders [13–16], suicidality [17–19], post-traumatic stress disorder (PTSD) [15,16,19–23], and neurodegenerative disorders [14,24–26]. The US Veteran’s Administration routinely screens Veterans for blast exposure and/or head injury to better understand the relative impact of TBI on long-term health-related outcomes [27]. Part of the difficulty in identifying long-term effects of TBI is the heterogeneity in presentation of symptoms and underlying pathology (i.e. differential circuits affected across individuals depending on the nature of the injury), with a need for objective, quantifiable markers of injury and risk for long-term dysfunction.

Some TBI patients display chronic sensory and information processing difficulties [28–34] which can disrupt a myriad of cognitive and emotional functions. Sensorimotor gating is a pre-attentional filtering mechanism which protects and adjudicates processing of sensory information [35–38]. Sensorimotor gating is operationalized via prepulse inhibition (PPI), a phenomenon in which responses to loud startling stimuli are attenuated if preceded by a weak pre-stimulus 30-300 msec before the startling stimulus. This short interval in which the prestimulus gates responses to subsequent stimuli preserves the processing of the initial prestimulus. Previous studies have identified sensorimotor gating deficits as a common biomarker and/or risk factor for several neuropsychiatric disorders [39–41], including bipolar disorder, schizophrenia, obsessive compulsive disorder, Tourette’s, and PTSD. PPI is mediated by a brainstem circuit modified by both cortical and subcortical circuits, making PPI a potentially sensitive marker of different neural circuit dysfunction and heterogeneous injuries [42].

In animals, PPI is disrupted across multiple TBI types and severity levels, and is sensitive to interventions [43–50]. Animal models of PPI are a translational measure of sensorimotor gating and pre-attention processes in humans, with face, construct and predictive validity (e.g. predictive of treatments that can modify PPI/associated symptoms in humans) [59–62]. PPI is an automatic process that is insensitive to language, motor, motivational or other factors confounding neurocognitive testing in humans. Given these attributes, PPI can be a useful translational, quantitative, objective marker of TBI-associated disruption and recovery. However, no well-controlled and powered studies have investigated the effects of TBI on PPI in humans, although small studies have investigated effects on other measures of sensorimotor gating (N=10-20) [51,52]. Although these studies suggest TBI may be associated with sensorimotor gating disruptions in humans, their small sample sizes and limited data on PPI and TBI injury variables hinder interpretability of these results.

Here we tested the hypothesis that TBI is associated with reductions in sensorimotor gating as measured by PPI, moderated by TBI severity. We used data collected from the Marine Resiliency Study (MRS), a study of active-duty Marines and Navy Corpsman preparing for a deployment to Iraq or Afghanistan (for study details see [53]), to examine lifetime TBI history associations with sensorimotor gating performance. Participants completed a suite of 3 acoustic startle tests, assessment of startle threshold [54] assessment of affect modulation of startle [55] and a PPI task [41]. Participants also completed symptom self-report instruments and clinical interviews, including detailed assessment of TBI history. We examined whether history of TBI is associated with sensorimotor gating as measured by PPI or habituation to acoustic startle. Finally, we analyzed whether TBI-type, severity, number of TBIs and recovery time modify TBI-associations with startle, habituation, or PPI.

Methods and Materials:

Study Design and Participants:

Data was extracted from the MRS, which included 4 infantry battalions from 2008-2012. After obtaining written informed consent, 2560 participants were evaluated for PPI approximately 1 month before a 7-month deployment to Iraq or Afghanistan. Detailed demographics for the MRS sample have previously been described and published [53]. At the time this study was conducted, women did not serve in infantry battalions, therefore all participants were male. Average age at the time of data collection was 22.8 years (Demographics shown in Table 1). Participants underwent mental fitness testing (ASVAB) and psychiatric illness screening prior to active-duty service enrollment. Approval for this study was obtained from the institutional review boards of the University of California, San Diego, the Veterans Affairs San Diego Research Service, and the Naval Health Research Center.

Table 1:

Participant Demographics. N.A. = Native American.

	No TBI	Mild TBI	Mod/Sev TBI	Overall
N total	773	824	128	1725
Age (yrs)
Mean	22.6	23	22.5	22.8
Range	18-45	18-43	18-37	18-45
Education (%)
< High School	2.2%	2.5%	3.1%	2.4%
High School	67.5%	61.4%	60.2%	64.1%
Some College	25.4%	31.4%	30.5%	28.6%
Bachelors	4.3%	3.8%	4.7%	4.1%
Post-Graduate	0.4%	0.8%	0.8%	0.6%
Missing Data	0.35%		0.8%	0.2%
Ancestry (genetic)
African	4.8%	2.9%	1.6%	3.7%
Asian/Other	13.1%	10.4%	7%	11.4%
European	61.7%	68.4%	75%	65.9%
Hispanic/N.A.	20.4%	18.2%	16.4%	19.1%
# of TBIs
1		47.8%	25.8%	24.7%
2+		52.2%	74.2%	30.4%
Deployment TBI		17.1 %	15.6 %	9.3%
Months since TBI
Mean (SD)		56.7 (57)	53.1 (49.8)	56.2(56)
Median		37	38	37
CAPS Totals
Mean (SD)	9.6 (10.2)	13.4 (11.3)	17.9 (12.2)	12.0 (11.2)

Exclusions:

Of the 2560 participants evaluated, 14 were excluded for hearing problems [41]. Since PTSD diagnosis is associated with low PPI performance [41], 187 participants were excluded for meeting the Diagnostic and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) diagnostic criteria for PTSD at the time of data collection (note alternative analyses using PTSD as a covariate with these participants in the models yielded similar results and effect sizes). Finally, 634 participants were excluded due to low startle responding (i.e. poor signal/noise based on predetermined cutoff, see [41]) preventing reliable measurement of inhibition of startle, leaving 1725 participants for analysis. Supplemental Table 1 shows a demographic breakdown of the initial vs final sample. Importantly, the proportion of key variables (endorsement of TBI, TBI type, demographics) were comparable across the full and final datasets, except for a lower proportion of participants with African ancestry in the final sample.

Measures:

The MRS conducted biological, physiological and psychosocial assessments including clinical assessments of TBI history. Genotyping was conducted [56] and is used for assessment of ancestry in this analysis rather than self-reported race/ethnicity. For this study, only measures used in the present analyses are described in full (for all measures collected see [53]).

TBI History Classification:

The Department of Defense (DoD) definitions of TBI and injury stratification were used to define TBI groups. The DoD defines a TBI as a head injury which results in an altered mental state or loss of consciousness (LOC), no matter how brief [4]. Participants reported up to 5 head injuries which resulted in “some alteration in consciousness”, including confusion, “seeing stars”, or a total LOC [57] and details including TBI-type, acute symptoms, severity, and date. If they endorsed more than 5 injuries, they were asked to describe the 5 worst events (see Table 1). For participants reporting multiple head injuries, their most severe TBI event was used for grouping.

Clinician Administered PTSD Scale (CAPS):

Because PPI is a risk factor for PTSD (see [41]), we assessed participants’ PTSD symptoms for possible exclusion or covariation. A partial PTSD diagnosis at the time of PPI collection was determined via a structured diagnostic interview (CAPS, [58]) using DSM-IV criteria: a threat of injury, physical integrity, or to life (Criterion A1), ≥ 1 re-experiencing symptom, ≥ 2 avoidance symptoms (standard is 3, lenient criteria used), and ≥ 2 hyperarousal symptoms [59–61]. Participants must have reported symptoms at least once within the past month (frequency ≥ 1) which caused a moderate to severe amount of distress (intensity ≥ 2) for a PTSD diagnosis [62]. Assessors had high inter-rater reliability for CAPS scores (intraclass correlation coefficient = 0.99 and PTSD diagnoses Kappa = 0.714; see [53].

Stimuli and Apparatus:

Detailed methods for MRS electromyography (EMG) collection were published previously: [41,63]. SR-HRLAB (San Diego Instruments, San Diego, CA, USA) delivered acoustic stimuli and recorded EMG signals. EMG signals were amplified (0.5mV electrode input=2500-mV signal output), band-pass filtered (100-1000Hz), digitized and recorded (1kHz sampling rate, all electrode impedance<10kΩ).

Psychophysiology experimental procedures:

Hearing was scored as detection of a 35-dB tone across both ears at multiple frequencies (500, 1000, and 3000 Hz) using a Grayson-Stadler Audiometer (score range 0-6). Previous analyses indicated that PPI performance is reduced in participants that detect fewer than 3/6 tones, thus individuals scoring <3 (n=14) were excluded (see [41]). After exclusions, no significant correlation between PPI performance and hearing score was detected (Pearson’s r=−.003, NS).

During EMG testing participants wore headphones and had two electrodes (Ag/AGCl) placed 0.5 cm apart at the left orbicularis oculi muscles and a reference electrode over the left mastoid bone. EMG sessions included a 5-minute acclimatization period to a 70-dB broad-band background noise, then a suite of 3 acoustic startle tests: startle threshold [54], affect modulation of startle [55] and PPI [41] in that order with 3-minute breaks in between. Only startle habituation and PPI trials are presented here.

The PPI session consisted of a 90 sec acclimation period before presentation of multiple trial types. “Pulse only” trials consisted of a 40ms, 114dB startle pulse, while PPI trials consisted of an 86dB 20ms prepulse followed 30, 60 or 120ms later (varied interstimulus interval) by a 114dB 40ms pulse. The average inter-trial interval was 15 seconds (range 7-23sec). Three 114dB pulse only trials were presented at the session start (first startle block), followed by a pseudorandom mix of ten 114dB pulse only trials (PPI startle block) and 18 PPI trials (six each of three trial types). Sessions ended with three 114dB pulse only trials (last startle block).

Startle habituation was calculated with the following equation:

$$\%\text{Habituation}=[\frac{firststartleblock-laststartleblock}{firststartleblock}X100]$$

Prepulse inhibition was calculated using the following equation:

$$\%\text{PPI}=100-[\frac{peakEMGresponsetoprepulsetrail}{peakEMGresponsetopulseonlytrialsduringPPIstartleblock}X100]$$

Data Processing:

Data were processed as previously described [41]. EMG data were smoothed (5ms rolling average) and a trained-rater blinded to groups removed artifacts (e.g. voluntary blinks, movement). Startle response was the peak EMG response within 100ms after pulse onset. Trials with excessively high noise levels that prevented startle response detection were removed [36]. Participants with average startle response to pulse alone <4 times baseline EMG signal (i.e., no-stimulus response; <4 signal/noise ratio) were removed to avoid floor effects of startle reactivity and/or excessively high noise confounding measurement of %PPI, see [41].

Statistical Analysis:

To investigate effects of TBI on PPI across the three ISIs, repeated measures ANOVAs were conducted. ISI was used as a within subject factor and TBI group (i.e. TBI history, TBI severity or LOC) as between subject factors. To assess effects on startle and %habituation, univariate ANOVAs were performed with participant scores as the dependent variable and TBI group as fixed factors. Supplemental Table 2 and Supplemental Table 3 contain all statistics, group means, standard errors, and Ns for startle assessments across TBI variables. In the case of significant main effects or interactions, post-hoc analysis was performed via Dunnett’s 2-sided t-test, with the “No TBI” controls as the reference group. For analyses of two continuous variables, bivariate correlations were performed with Pearson’s coefficient and p values reported in text. Effect sizes are reported via Cohen’s d or partial eta squared (ηp²) as appropriate. All statistical analyses were conducted using IBM SPSS Statistics (RRID:SCR_016479).

PPI data were checked for normality with no violations (kurtosis < |3| and skew < |2|). To address extreme outliers data were winsorized with scores >3 standard deviations from the mean assigned the standard deviation=3 score.

Potential covariates:

Potential covariates were examined for associations with PPI via Chi Squares or ANOVAs. Stimulant use: There were no associations of self-reported caffeine and nicotine with PPI [41,64]. Genetic ancestry: Participants were grouped into 4 GWAS-based categories: African, Asian/Other, European, and Hispanic/Native American (see [65], supplemental methods for details; Supplementary Table 1 and Table 1). Ancestry differed across TBI groups (Χ²(6, n=1725)=16.238, p=0.013) although there was no main effect of ancestry (F(3,1713)=1.038, p=NS) or interactions with ISI (F(6,3426)<1, p=NS), or TBI severity (F(6,1713)<1, p=NS) on PPI, in line with past studies [41,66,67]. Hearing: After exclusions, there was no significant relationship between hearing and any variable presented here (Fs<1.61, ps>0.20) except for blast TBI exposure (F(2,1721)=6.01, p=0.003). Post-hoc sensitivity analyses were also performed using number of tones detected as a covariate with no change in findings (pulse alone startle, habituation, PPI). CAPS: To complement the conservative approach of excluding all participants with a current PTSD diagnosis, we conducted sensitivity analyses with CAPS scores as a covariate (continuous scale, 0-136), finding no differences in TBI effects and it was dropped from final models. Hence all covariates above were dropped from final models.

Results

Demographics:

See Table 1 for breakdown of demographics across TBI groups. TBI severity groups did not significantly differ in age (F(2,1722)=2.33, p=.098) or education (Χ² (8, n=1722)=10.195, p=NS). Fifty-two percent of participants who reported a TBI reported more than one. Over 80% of participants received their last TBI >1 year before data collection, with median time since injury being 56 months prior to data collection. Participants were majority European ancestry (62-75%), with Hispanic/Native American being the largest minority represented (16-20%). Genetic ancestry significantly differed across TBI groups, an effect driven by fewer Europeans in the no TBI control group (Χ² (6, n =1725)=16.238, p=0.013)).

TBI history:

When grouping participants into those that have or have not been exposed to at least one DOD-defined TBI there were no significant differences in PPI performance between groups (Figure 1A, Main effect of TBI: F(1,1723)<1, p=NS, ISI*TBI: F(2,3446)<1, p=NS). Startle reactivity and habituation did not differ across groups (F(1,1723)<1, p=NS; F(1,1715)<1, p=NS respectively).

Figure 1: Effects of Previous TBI on Prepulse Inhibition.

A: Lifetime TBI History (yes/no). Ns: No TBI=773, TBI=952. B. Severity of previous TBI. Ns: No TBI=773, Mild TBI =824, Mod/Severe TBI=128. *p<0.05, Main effect of TBI Group. C. Duration of LOC with previous TBI. Ns: No LOC or TBI=1114, 0<LOC<30min=535, LOC>30min=76. * p< 0.05 Dunnet’s test vs control (no TBI) following significant interaction between ISI and TBI group. D. TBI with or without Blast Exposure, Ns: No TBI=773, Blast TBI=182, Non blast TBI=770

TBI severity:

We then investigated effects of severity of TBI. Participants with DOD definition of moderate-severe TBI exhibited significantly lower PPI performance compared to the No TBI group (Figure 1B, Main Effect of Severity F(2,1722)=4.245, p=0.014, ηp²=.005, Severity*ISI: F(4,3444)<1, p=NS). Dunnett’s post-hoc tests revealed participants with previous moderate-severe TBIs had significantly lower PPI compared to controls (p=0.029, d=.23). TBI severity was not associated with differences in baseline startle reactivity (F(2,1722)=1.701, p=NS) or startle habituation (F(2,1714)<1, p=NS).

We next asked what TBI severity variables might best explain the association given that both PTA and LOC intervals contribute to TBI severity stratifications. TBI with a LOC>30 min is significantly associated with lower PPI scores, which was dependent upon ISI (Main effect of LOC: F(2,1722)=3.285, p=0.038, ηp²=.004, LOC*ISI: F(4,3444)=2.556, p=0.037, ηp²=.003, Figure 1C). Participants with LOC>30 min had significantly lower PPI compared to controls (p=0.039, d=.29) while participants with LOC<30 minutes did not (p=.723, d=.02). Participants with LOC>30 min had significantly lower PPI compared to controls specifically in the shorter ISI trials of 30 and 60 msec (Fs>3.68, ps<0.026, .31>ds>.29). Baseline startle and habituation were not different between LOC groups (all Fs<1.4, p=NS, see Supplemental Table 2)

Sensitivity analyses indicated that LOC group effects remained when covarying for PTA, the other key criteria used for DoD severity definitions (Main effect of LOC: F(2,1721)=3.54, p=0.029, ηp²=.004, LOC*ISI: F(4,3442)=2.709, p=0.029, ηp²=.003). PTA alone was not significantly associated with PPI (Main Effect of PTA: F(2,1722)=2.428, p=0.089, PTA*ISI: F(4,3444)<1, p=NS), startle reactivity or habituation (startle: F(2,1722)=1.358, p=NS; habituation: F(2,1714)=.565, p=NS).

Blast TBI:

Because blast TBI has a specific set of injury sequelea, we examined whether exposure to blast TBI alters sensorimotor gating. Blast-specific TBI was not associated with altered PPI performance compared to non-Blast TBI and No TBI groups (Figure 1D, Main Effect of Blast: F(2,1722)<1, p=NS, ISIxBlast: F(4,340)=1.047, p=NS). Startle reactivity across sessions decreased across all groups as expected (Figure 2A, Main Effect of Startle Trial: F(1,1714)=1725.323, p<0.001, ηp²=.502). However, there was a trend for a main effect of blast group and interaction of blast group with startle trial (Figure 2A, Main Effect of Blast: F(2,1714)=2.763, p=0.063, ηp²=.003, Blast * Startle: F(2,1714)=2.938, p=0.053, ηp²=.003). Blast TBI was associated with lower startle habituation compared to controls (Figure 2B, F(2,1714)=3.566, p=0.028, ηp²=.004; Dunnett’s test blast TBI compared to controls p=0.038, d=.19). Sensitivity analyses were conducted to examine if hearing loss or higher initial startle account for the reduced habituation in the blast TBI group. Significant differences remained when using hearing score and initial startle as covariates (Main effect of Blast-TBI, F(1,1711)=4.197, p=0.015, ηp²=.005).

Figure 2: Effect of Blast Exposure on Acoustic Startle and Habituation.

A: Startle reactivity across session duration: Startle reactivity to first startle block (left) and last startle block (right) to 114 dB pulse. B. Percent habituation of startle over entire EMG testing period: (first block – last block)/first block * 100). Error bars = Group Mean +/− 1 SEM. *** = p<0.001, * = p<0.05 vs control (No TBI). Ns: No TBI=773, Blast TBI=182, Nonblast TBI=770.

Time since TBI, number of TBIs, and current symptoms:

Additional modifying variables were explored to understand if PPI performance changed with time since injury, number of TBIs and self-reported TBI symptoms, with none of the variables accounting for significant variance in PPI (see Supplemental Results). Startle habituation, but not startle reactivity or PPI, was associated with increased self-reported symptoms, with participants endorsing 4+ symptoms (out of 6) exhibiting significantly reduced habituation compared to controls (F(2,1714)=6.478, p=0.002, ηp²=.008; see Supplemental Figure 2 for details and analysis approach).

Discussion

This study examined lifetime TBI exposure associations with sensorimotor gating performance. Participants who endorsed a moderate-severe TBI exhibited lower PPI scores compared to controls. Of the defining features of TBI severity, only LOC duration predicted PPI performance, while PTA did not. We did not observe significant effects of time since TBI, multiple injuries, or symptoms associated with TBI on PPI performance. While startle habituation was not affected by TBI severity, blast injury and increased number of TBI symptoms endorsed was associated with lower startle habituation. No group exhibited disrupted baseline startle and findings survived correction for hearing, indicating the sensorimotor gating disruptions were not due to baseline startle or hearing loss. These results suggest that loss of consciousness and injury severity are the primary features affecting TBI associations with low PPI performance, rather than number of head injuries or endorsement of current symptoms. These data indicate that injury severity is the most likely predictor of low PPI performance, and that these effects are likely long-term.

Mild TBI was not associated with lower PPI, despite prior reports of attentional and functional disruption in humans [28–34]. One possible explanation for detecting only moderate-severe effects may be due to the injury timeline in this dataset. Most mild TBI patients experience recovery across cognitive domains, with most impairment resolving within a 3-12 month window after injury [68]. Within our cohort, over 80% of participants obtained their most recent injury >1 year ago (median=37 months, mean=56 months). This long interval may limit the ability to detect time-dependent functional effects on PPI, since acute effects may have resolved before testing. As such, these associations likely reflect chronic reductions in sensorimotor gating after a moderate-severe injury. Indeed, animal models of mild TBI often result in a rapid functional recovery window. In rats exposed to repetitive mild blast injuries (7.5-25 PSI), acute deficits are observed across a wide range of blast severity parameters, however only the strongest blast events produced PPI deficits that remained >1 week [46]. Rats given a single mild blast TBI (14 PSI) had significant reductions in PPI one day after exposure, followed by recovery at assessments 14, 28 and 90 days post injury [44]. Similarly, mild TBI induced via controlled cortical impact disrupts sensorimotor gating acutely, with symptoms resolving by week 6 follow-up [47]. These findings support the idea that like most functional impairments after mild head injury, sensorimotor gating can recover in the weeks following injury. Future longitudinal investigations in humans after acute injury are needed to resolve the pre-attentional sensorimotor gating disruption, risk, and recovery timeline.

PPI is categorized by both “automatic” and “controllable” components dependent on ISI duration, with ISI>100ms thought to be modifiable via conscious attentional control [69] while shorter ISI parameters probe automatic, unconscious components. LOC associations with PPI depended on ISI, with TBI at 30-60ms ISI associated with reduced PPI and no difference at 120ms. These results suggest that TBI with a LOC>30min is associated with reduced PPI specifically during the “automatic” processing window. Interestingly, these findings resemble short-ISI PPI deficits observed in other neuropsychiatric disorders, including schizophrenia and PTSD [41,66]. Thus, neurological mechanisms which underlie automatic pre-attentional filtering performance may be more sensitive to disruption with TBI, in line with parameters most associated with neuropsychiatric risk.

PPI is often used to measure underlying neurological dysfunction and serves as a biomarker of and/or risk factor for the development of neuropsychiatric disorders including schizophrenia, bipolar disorder, mood disorders, PTSD, Tourette’s, and OCD [37,41,70–74]. Injuries which disrupt sensorimotor gating processing may confer higher risk for the subsequent development of neuropsychiatric conditions, or PPI may serve as a marker of shared circuit disruptions rather than mediate TBI-associated risk for neuropsychiatric disorders per se [15]. The current cross-sectional study cannot address the causal relationship between TBI, sensorimotor gating and development of neuropsychiatric symptoms. An alternative explanation to TBI disrupting PPI is that participants with lower sensorimotor gating could be predisposed to behaviors which make them more at risk for TBI exposure. However, human and animal studies suggest high impulsivity is not associated with PPI performance [75–78]. Additionally, if lower sensorimotor gating increased impulsivity and TBI-related accidents, a relationship with TBI severity would not be expected. Instead, we would expect to observe a relationship between PPI and number of lifetime TBIs sustained, which was not detected. However, longitudinal analysis is needed to fully rule out this alternative hypothesis.

Although blast TBI exposure was not associated with PPI performance, it was associated with reduced startle habituation. Like PPI, reduced startle habituation is detected across a number of neurodegenerative and neuropsychiatric disorders, including Parkinson’s disease, Huntington’s disease, PTSD, schizophrenia, and schizotypal personality disorder [79–85]. Startle habituation also involves spinal, subcortical and cortical circuits [86], suggesting that like PPI, it is a sensitive but non-specific measure of brain function with multiple potential points of dysfunction depending on the focal point of injury. Reduced startle habituation may be reflective of unique pathology or sequalae underlying blast TBI. Unlike non-blast TBIs, primary blast injuries result from explosive biokinetic energy overpressure waves penetrating into tissues which trigger an injury cascade, resulting in the compression, shearing, vascular damage, diffuse hemorrhaging, and widespread injury of soft cerebral tissues [87–91]. This differs from non-blast injuries, which typically involve direct focal damage at the site of impact [88], suggesting blast may be more likely to disrupt additional functions that contribute to sensory habituation compared to focal head injuries. Because of the diffuse nature of TBI, regional effects are heterogeneous. However in animal models of TBI with PPI and/or acoustic startle disruption, these effects appear to be due to either overt neuronal loss [44], axonal degeneration [43] or disruption of cell function via metabolic or inflammatory pathways [92–99]. Overall, PPI and startle habituation serves as sensitive and highly conserved measures of neurological dysfunction associated with TBI and may prove useful in identifying and validating TBI interventions across species.

Study Limitations

A limitation of this study is that the population is homogenous across age (range 18-45 years, >80% subjects below 25 years, >90% all subjects below 27 years), ancestry (primarily of European), occupation (service members) and sex (only males). Women were not included in this study due to service restrictions in Marine infantry divisions. Sex can influence sensorimotor gating and TBI outcomes [100–104] indicating more research is needed to test the generalizability of these findings to different populations. Additionally, a relatively high percentage of subjects were lost due to low signal to noise levels (<4x, 24.7%). Since data presented here were collected during a field study in sequence with other assessments, participants had greater exposure to sweat, dirt, or other environmental factors which we suspect contributed to the high signal to noise ratio loss. Despite these limitations, this study has several key strengths including: the large study cohort (N=1725), parametric assessment of PPI with ISIs, clinical interviews to establish DoD diagnostic TBI criteria/stratification, and controlling for confounding PTSD symptomology and hearing. It is important to note that although this study is in service members, 83% of TBIs were non-deployment related, making findings potentially more relevant to civilian injuries.

Conclusion:

These findings support the hypothesis that prior moderate to severe TBI exposure and blast injuries are associated with long-term decreased sensorimotor gating, a previously validated metric for neuropsychiatric risk. Future studies should aim to investigate acute recovery windows for this observed effect, sex differences, and civilian participants. To make causal inferences and understand impacted circuitry driving TBI-associated sensorimotor gating disruptions, future neuro-mechanistic studies, prospective longitudinal studies, and imaging investigations should be prioritized.