A Closed-Head, Rotational Traumatic Brain Injury Model demonstrates Deficits in Righting Reflex, Neurological Function, and Auditory Brainstem Responses in Rats

Abstract

AIMS

The pathophysiology of traumatic brain injury (TBI) is incompletely understood and warrants further investigation with preclinical models that recapitulate aspects of the human condition. Therefore, our goal was to quantify the initial presentation of deficits in a closed-head rotational TBI model in male and female rats in return of righting reflex (RORR) and global neurological function including gross sound reflex scores as a measure of acoustic startle.

METHODS

A total of 116 rats (43 female) were included in this study that underwent TBI (n = 53), Sham (n = 42), or No Anesthesia control (n = 21) procedures to assess RORR, the Revised Neurobehavioral Severity Scale (NSS-R), or Auditory Brainstem Responses (ABR).

RESULTS

TBI rats had a trend towards a prolonged RORR, which was significantly increased in the TBI compared to control group when normalized to the duration of anesthesia. Gross neurological function was impaired in TBI rats, such that descriptively the largest component was due to acoustic startle deficits. These deficits were not explained by worse (higher) hearing thresholds in auditory brainstem responses, although TBI rats had increased thresholds at 16 and 20kHz, and a trend at 10kHz compared to Shams on post-TBI day 1 that was not different by day 8.

CONCLUSIONS

This model reproduces acoustic startle, and hearing impairment aspects of human TBI that influence quality of life. Additional studies are necessary to integrate variables from new clinical classification systems including blood-based and imaging biomarkers to aid in the translation to improve patient-oriented outcomes.

Introduction

Traumatic brain injury (TBI) care has greatly improved over time (1), but the pathophysiological mechanisms are incompletely understood, which has limited advances in patient-oriented outcomes in recent years (2). A better understanding of TBI would benefit military service members, civilians, and their families by improving prognostication of functional ability, mental health (3), cognition (4, 5), and quality of life (6), which is currently limited by the various recovery trajectories that occur despite similar initial clinical presentations (7–9). Animal models are necessary to recapitulate various aspects of TBI in humans (10), including on the more severe end of the injury spectrum (11). Modeling human TBI with animal surrogates using rotational injury devices has become more prevalent in recent years (10) due to the ability to more accurately correlate translatable biomechanical measures such as head rotational acceleration with resulting functional deficits as a result of diffuse brain injury rather than focal deficits (12, 13). Animal models that generate more severe diffuse injury are generally lacking since current models often have a focal injury component (11). Biofidelic modeling of distinct TBI severities is essential because secondary insults might depend on the pathoanatomical subtype. For example, we have recently shown that exposure to extracranial surgery/anesthesia shortly after TBI was associated with worse brain injury specific outcomes (14, 15). However, the effect was limited to patients with moderate to severe decreases in alertness or responsiveness as quantified with Glasgow Coma Scale (GCS) or TBI that had a computed tomography (CT) scan demonstrating intracranial pathology (e.g. bleeding) despite a Glasgow Coma Scale (GCS) score of 13–15 consistent with preserved responsiveness (14, 15). Therefore, we exposed rats to a previously reported Rotational Injury Device (RID) that elicits injury to diffuse brain regions (12, 16–19), but we targeted injuries with moderate acute deficits in alertness and responsiveness.

The objective of this study was to establish the initial presentation of a novel closed-head rotational TBI rat model targeting a prolonged loss of responsiveness and moderate neurological dysfunction. Brain mass scaling between animals and humans (20) was used to approximate the biomechanical parameters necessary to increase TBI severity from previously published results using this coronal plane RID model that generated subconcussive (21), classical concussion, or “mild” TBI including loss of consciousness determined as an extended time to responsiveness following anesthesia (12, 16–19). Our approach to quantifying the neurological dysfunction induced was to grossly identify the deficits with the largest effect sizes. This was accomplished by implementing the Revised Neurobehavioral Severity Scale (NSS-R), (22) which is a brain injury scoring system that has been validated in Blast (23), fluid percussion (24), controlled cortical impact (25), impact acceleration (26), and weight drop TBI models (27). The NSS-R was selected because it has been validated in many TBI models, has not been tested in a coronal plane rotational model to our knowledge, and it includes 10 different tests that quantify rodent behaviors using a gross scoring system assigned a score of zero (healthy), one (compromised), or two (absent). The hypothesis was that this rat TBI model would replicate aspects of human TBI, including loss of consciousness, weight loss, and neurologic dysfunction.

Methods

Animals

For these studies, 132 adult Sprague-Dawley rats, 82 males (300–325g) and 50 females (225–300g), were ordered at 9–14 weeks-old. Euthanasia endpoint criteria due to TBI included respiratory distress, severe neurologic signs such as seizure activity lasting more than 10 minutes continuously, severe ataxia, marked depression of activity, lack of movement or minimal response to tactile stimulation, or other signs consistent with a comatose state. Data collected from animals euthanized due to TBI-related endpoint criteria were included in the analysis if available, and unavailable data were treated as missing. There were 16 mortalities unrelated to TBI exposure. All animal procedures were approved by the Institutional Animal Care and Use Committees prior to the initiation of experiments (Supplemental Materials).

Traumatic brain injury

The RID administered TBI (n = 53 rats) with the goal of inducing moderate acute functional neurological deficits without including debilitating injuries based on the euthanasia criteria above. Briefly, this model implements a closed head, coronal plane rotational injury without surgical interventions by securing the rat’s head into a helmet, which is rotated when an impactor rod strikes a moment arm attached to the helmet by a beam that is the center of rotation (12, 16) (Supplemental Materials). An accelerometer attached to the helmet was used to calculate TBI biomechanics, including the peak rotational acceleration magnitude, duration, and integral (area under the curve; AUC) of the rotational acceleration and deceleration curves (Figure 1). Recapitulation of human TBI biomechanics in rodent models requires brain mass scaling (20) so female rats were intentionally allowed to be an older age to better match weights such that TBI biomechanics were more similar. This approach minimized the influence of brain size but introduced age as a potential confound. In pilot studies, TBI administered targeting 600 krad/s/s impacts each up to 5 impacts often displayed subtle clinical findings or no phenotype (weight loss, piloerection, porphyrin staining, or gross neurologic changes). Therefore, the RID was targeted at 800 krad/s/s per impact, which is the maximum settings able to achieve consistent biomechanics with the current design. In a rat with an approximately 1.8 gram brain mass, an 800 krad/s/s impact is roughly equivalent to 9.6 krad/s/s rotational injury in a human, which is consistent with an Abbreviated Injury Scale (AIS) of 3 or an Ommaya Gennarelli Concussion Grade of 4 that would be predicted to generate loss of consciousness for an hour (20). The experiments were planned for 10 impacts. Since the euthanasia endpoint criteria were only observable after emergence from anesthesia, the number of impacts was decreased if there were TBI-induced vital sign derangements (heart rate less than 300 or greater than 600 bpm, blood pressure less than 90 or greater than 200 mmHg, or oxygen saturation less than 90%), which were anecdotally associated with reaching euthanasia criteria. All impacts were administered in as rapid succession as possible given the limitations of reloading the impactor rod and restarting the accelerometer recording. TBI biomechanical parameters were quantified using area under the curve (AUC) in order to assess for outcome differences due to small variations between the animals in this study, which would not be valid across a wider spectrum of injury severities. For example, it is well accepted that a TBI with a single impact of 8,000 krad/s/s is not equivalent to 10 impacts of 800 krad/s/s per impact, despite both having an equivalent cumulative peak magnitude.

Figure 1. Rotational Injury Device Biomechanics.

Representative tracing of rotational acceleration peak magnitude, duration, and area under the curve yielded from a rod ejected at the moment arm of the traumatic brain injury device.

Prior to the RID procedure, 5% isoflurane was administered in a chamber. A surgical plane of anesthesia, characterized by immobility in response to noxious stimuli, was established prior to endotracheal intubation. All TBIs were performed under the same anesthetic and therefore only a single return of righting reflex (RORR) was recorded. The time to the RORR was recorded as the time from turning off the volatile anesthetic to the transitioning from supine to prone. Raw RORR was divided by the duration of anesthesia and reported as the normalized RORR. We subsequently dichotomized the TBI group based on RORR times greater than (prolonged) or less than (normal) 2 SD greater than the mean of the Sham group values.

Control Groups

A Sham group (n = 42 rats) was subjected to the entire procedures, including general anesthesia, intubation, and insertion into the TBI helmet, without exposure to rotational acceleration of the head. A “No Anesthesia” group (n = 21 rats) did not experience any exposure to anesthesia throughout the course of the study, so animals were neither inserted into the RID helmet nor intubated. This is a critical control group that should be included in studies that implement isoflurane since anesthetic-induced neurotoxicity has been documented in rodents (28).

Neurological Function

The NSS-R was used to quantify deficits in global neurological function at baseline and on post-TBI days 1, 4, and 8 (22). The NSS-R is a series of specific, sensitive, and standardized observational tests that evaluate balance, motor coordination, and sensorimotor reflexes in rodents (Supplemental Materials) (22–25). The rater was blinded to the injury status of the rats.

Auditory Brainstem Responses

Recordings of Auditory Brainstem Responses (ABR) were performed in rats anesthetized with ketamine (75mg/kg; i.p.) and dexmedetomidine (0.5mg/kg; i.p.) since isoflurane is known to dampen evoked potentials. Dexmedetomidine was reversed after the procedure with atipamezole (0.5–2.5mg/kg; i.p.). Each rat was tested at 10, 16, and 20kHz, such that each frequency started at 90dB and decreased by 5dB until reaching the hearing threshold, or the minimum testable intensity of 20dB. Additional details available in Supplemental Materials. Tympanic membranes were not evaluated (29). The hearing threshold was determined as the lowest stimulus intensity at which wave IV was present and repeatable to visual discrimination by a single trained experimentalist (Figure 2) (30).

Figure 2. Auditory Brainstem Response Testing.

(A) Waveforms of electrophysiological recordings are shown for a representative animal demonstrating decreased amplitude of wave IV with sequential testing of 10kHz at lower decibel (dB) settings until complete loss at 25dB. The hearing threshold for this animal is therefore 30dB. (B) All animals were tested in 5dB increments and threshold was verified by reproducibility (with 3 recordings) as shown at increased amplification.

Statistical Analysis

Control animals (Sham, No Anesthesia) were compared using one-way ANOVAs for age, baseline weight, maximum weight loss, and NSS-R to assess if groups could be combined. Post-tests were performed using Tukey.

A Mixed Linear model (PROC MIXED) was used to assess differences in NSS-R between TBI and control groups accounting for sex, age at time of TBI, baseline weight, duration of anesthesia, normalized RORR, and time with animal as the random effect. A separate mixed model assessed NSS-R using all the same covariates above, but additionally included number of TBI impacts, cumulative TBI magnitude, cumulative TBI duration, and cumulative TBI AUC to determine if an individual biomechanical parameter predicted NSS-R deficits. Three separate General Linear models (PROC GLM) were used to assess differences loss between TBI and control groups in RORR, total NSS-R, and maximum weight. These models controlled for sex, age at time of TBI, baseline weight, number of TBI impacts, cumulative TBI magnitude, cumulative TBI duration, cumulative TBI AUC, duration of anesthesia, and time as fixed effects. The two separate models for total NSS-R and weight loss also included normalized RORR as fixed effects.

Total NSS-R scores are a summary score and therefore continuous data, but individual components of the NSS-R, such as sound reflex and balance scores are ordinal data. Consequently, these were analyzed using Conditional Logistic regression (PROC LOGISTIC) and Chi Square test, respectively. ABR thresholds were analyzed individually for each frequency tested using repeated measures ANOVA via a PROC GLM with the sexes collapsed due to the smaller sample size in those experiments.

A two-sided threshold of p<.05 was used for statistical significance. All analyses were conducted using SAS 9.4. Results are reported as mean+/−SD unless noted otherwise. Figures were generated with the ggplot2 package in R, Prism Graphpad 10.3.1, and Adobe Photoshop 2025.

Results

Subject Characteristics

On the day of TBI and control procedures, the 116 animals that remained in the study were 15+/−3.9 weeks old with 43 female (37%) and 73 male (63%) that were 296+/−42g. As expected, based on our methodological approach, there were sex differences in baseline weight and age, such that the females were lighter (p<0.0001) and older (p<0.0001) than the males (Table 1) according to one-way ANOVA. The control groups (Sham and No Anesthesia) were compared using one-way ANOVA and Chi Square tests to assess whether they were different from each other with the a priori plan to collapse groups if not different. No age differences were detected between the Sham or No Anesthesia groups in either females (18.3 +/− 3.9 weeks Sham; 18.8 +/− 4.0 No Anesthesia) or males (12.5 +/− 1.9 Sham; 14.2 +/− 2.2 No Anesthesia). No baseline weight differences were detected between the Sham or No Anesthesia groups in either females (18.3 +/− 3.9 Sham; 18.8 +/− 4.0 No Anesthesia) or males (12.5 +/− 1.9 Sham; 14.2 +/− 2.2 No Anesthesia). There were no differences between the control groups in maximum weight loss or NSS-R scores (Table 1) and therefore the control groups were combined for all subsequent analyses (Table 2).

Table 1.

Animal Characteristics.

Variable	Female									Male									p-value
	No Anesthesia			Sham			TBI			No Anesthesia			Sham			TBI
	n	Mean	SD	n	Mean	SD	n	Mean	SD	n	Mean	SD	n	Mean	SD	n	Mean	SD
Age at time of TBI (weeks)	9	18.8	4.0	16	18.3	3.9	18	18.3	3.9	12	9.5*	2.2	26	12.5*	1.9	35	13.0*	2.2	<0.0001
Baseline Weight (grams)	9	250	15	12	247	25	17	253	25	9	320#	11	12	325#	15	33	330#	15	<0.0001
Maximum Weight Loss (grams)	9	−3	4	15	−7	6	9	−16	9	12	−1	7	16	−5	10	23	−29	20	-
Total NSS-R
Baseline	5	0.0	0.0	10	0.0	0.0	10	0.0	0.0	6	0.3	0.5	21	0.1	0.4	29	0.4	0.8	-
Day 1	5	0.2	0.4	10	0.1	0.3	10	1.9	3.3	6	0.2	0.4	22	0.1	0.2	28	3.8	4.0	-
Day 4	5	0.0	0.0	5	0.2	0.4	5	0.8	1.8	2	0.0	0.0	12	1.0	1.1	12	3.0	3.3	-
Day 8	1	0.0	-	6	0.2	0.4	4	0.5	1.0	1	1.0	-	11	0.5	0.5	9	1.0	1.3	-

Main effect of sex reported from one-way ANOVA with post-tests that demonstrated that females were *significantly older than males, and ^#weighed more than males in the same treatment group. There were no differences between the No Anesthesia and Sham groups according to one-way ANOVA so those groups were combined into a single Control group for all subsequent analyses. Cells with a single dash denote analyses that were not performed with all groups separated since they were outcomes of interest.

Table 2.

Descriptive data for TBI and control procedures.

Variable	Female						Male						p-value
	Control			TBI			Control			TBI
	n	Mean	SD	n	Mean	SD	n	Mean	SD	n	Mean	SD
Duration of Anesthesia (min)	10	49	7.7	12	64	24	19	43	30	23	62	12	0.01 1
RORR (min)	10	7	2.5	12	20	28	19	7	4	23	15	11	0.067 2
Normalized RORR (min/min of Anesthesia)	10	0.14	0.05	12	0.34	0.51	19	0.13	0.10	23	0.27	0.23	0.03 3
Maximum Weight Change (grams)	24	−6	5	9	−16	9	28	−3	9	23	−29	20	0.14 4

Compared to the control group, the traumatic brain injury (TBI) group had a larger duration of anesthesia according to one-way ANOVA

¹a trend towards a longer time to return of righting reflex (RORR) that did not reach statistical significance according to a general linear model (PROC GLM)

²a larger (longer) normalized RORR according to a separate PROC GLM

³and a trend toward a greater maximum weight loss according to a separate PROC GLM.

⁴No sex differences were detected in any analyses.

TBI Biomechanics, Mortality, and Critical Injury Rates

The cumulative peak rotational acceleration magnitude from TBI was 9215 +/− 2650 krad/s², which corresponded to an AUC of 8290 +/− 2290 krad/s and a cumulative acceleration duration of 50.3 +/− 13.5ms over 8.7 +/− 1.6 injuries (Table 3). A total of 15 (13%) rats met euthanasia criteria (10 rats on the day of TBI; 5 rats on post-TBI day 1 or later).

Table 3.

Biomechanical parameters.

Variable	Female TBI (n = 18)			Male TBI (n = 35)			p-value
	Mean	SD	Coefficient of Variation	Mean	SD	Coefficient of Variation
Number of Impacts	8.6	1.5	17	8.8	1.7	19	0.57
Cumulative Magnitude of Impacts (krad/s/s)	8890	2760	31	9380	2610	28	0.53
Cumulative Duration (ms)	51	14	27	49	13	27	0.55
Cumulative AUC (krad/s)	7650	1990	26	8610	2390	28	0.15

TBI administration was not different between male and female Sprague Dawley rats according to one-way ANOVA.

Return of Righting Reflex

A General Linear model (PROC GLM) was used to assess RORR that controlled for TBI group, sex, age at time of TBI, baseline weight, number of TBI impacts, cumulative TBI magnitude, cumulative TBI duration, cumulative TBI AUC, and duration of anesthesia. There was a trend towards a difference between the TBI (17 +/− 18 min) and control groups (7 +/− 4 min) that did not reach statistical significance (p = 0.067; Table 2). There was no evidence of differences based on sex (p = 0.61), age (p = 0.97), weight (p = 0.23), or TBI parameters (number of TBIs, p = 0.61; cumulative TBI magnitude, p = 0.25; cumulative TBI duration, p = 0.18; cumulative TBI AUC, p = 0.29). A separate PROC GLM model was generated for RORR normalized to the duration of anesthesia, which showed the TBI group had an increased (longer) normalized RORR (0.29 +/− 0.34; p = 0.03) compared to the control group (0.13 +/− 0.086). There were no differences in sex (p = 0.65), age (p = 0.97), weight (p = 0.29), or TBI parameters (number of TBIs, p = 0.72; cumulative TBI magnitude, p = 0.24; cumulative TBI duration, p = 0.20; cumulative TBI AUC, p = 0.26). RORR data had a bimodal distribution and so was dichotomized based on normalized RORR times greater than (prolonged) or less than (normal) 2 SD greater than the mean of the Sham group values (Figure 3). In the TBI group, 12 (22.6%) had a prolonged RORR, of whom there were 9 (25.7%) male and 3 (16.7%) female. No animal in the sham group had a prolonged RORR.

Figure 3. Return of Righting Reflex.

The time to Return of Righting Reflex (RORR) was quantified in rats undergoing traumatic brain injury (TBI) and Sham procedures. The TBI group had a trend toward a longer RORR compared to Shams according to the PROC GLM (p = 0.067). Individual data points are presented with the median and interquartile range. Rats dichotomized into the prolonged RORR subgroup (defined as greater than 2 SD above the mean Sham group value) are illustrated by squares. A separate PROC GLM model was generated for RORR normalized to the duration of anesthesia, (data not shown; further details in Results; p = 0.03).

Body Weight Change

Weights were measured up to 5 days per week after TBI and the greatest change from baseline weight was recorded, regardless of the day on which it was observed. Maximum weight change differed by one-way ANOVA across sex-and-intervention (p < 0.0001), with the male TBI group losing more than female controls and with TBI-exposed animals of both sexes losing more than male controls. These effects of TBI did not hold after controlling for other factors using a GLM model, which found that maximum weight change was not explained by TBI exposure (p = 0.14) after accounting for sex (p = 0.59), age (p = 0.76), baseline weight (p = 0.17), TBI parameters (number of TBIs, p = 0.60; cumulative TBI magnitude, p = 0.53; cumulative TBI duration, p = 0.80; cumulative TBI AUC, p = 0.45), duration of anesthesia (p = 0.52), and normalized RORR (p = 0.48; Table 2). We also used a GLM to assess the total NSS-R in relationship to the maximum weight loss controlling for group, sex, age, baseline weight, TBI parameters, normalized RORR, and time. There was a relationship between maximum weight change and the total NSS-R score (p = 0.004), such that for every 10 grams of body weight loss there is an association with a 1.3 greater (more deficits) total NSS-R, which is the expected direction of change.

Neurological Function

Total NSS-R scores were increased (more deficits) in the TBI group compared to Controls according to a mixed model that found a significant effect (p < 0.0001) of day by group (TBI versus controls), while controlling for sex (p = 0.86), age (p = 0.75), and baseline weight (p = 0.50) (Figure 4A). The effect was limited to the TBI group on post-TBI day 1 (p < 0.01). All other groups and times were not different from controls at baseline (p > 0.05). The prolonged RORR group had a 1.7-point increase in total NSS-R score (more deficits) compared to subjects with a normal RORR following TBI (p = 0.015). In a separate mixed model, TBI parameters were included as covariates to determine their contribution to total NSS-R scores. This model also found a significant effect (p < 0.0001) of day by group (TBI versus controls), but no individual group or day was significant (p > 0.05). In this model, for every 1000 krad/s/s increase in cumulative TBI peak magnitude there was a 0.8-point increase in total NSS-R score (p < 0.01). The model also found a significant, but paradoxical effect of cumulative TBI AUC such that for every 1000 krad/s decrease there was a 0.5-point increase in total NSS-R score (p < 0.05). Similar to the initial model, this model also found no effect of sex (p = 0.86), age (p = 0.75), and baseline weight (p = 0.50).

Figure 4. Neurological Function.

Rats were assessed for (A) global neurological function using the total Revised Neurobehavioral Severity Scale (NSS-R), (B) acoustic startle using the sound reflex, and (C) vestibular system using balance which were tested at baseline, post-TBI day 1, 4, and 8 in the TBI, Sham, or No Anesthesia groups (not shown; no difference from Sham). Sound reflex scores quantify acoustic startle behaviors evoked by a 60–70dB hand clap that is approximately 10kHz. Balance scores assessed the ability to balance and walk on a beam. Columns represent percent in each group with a given score. Lower scores are better. Statistical significance in one-way ANOVA denoted for differences in the change from baseline between groups; **p<0.01; ***p<0.0001.

A major determinant of the total NSS-R scores was the sound reflex, which are ordinal data that violate the assumptions of the chi square test and therefore are presented descriptively. Control rats scored perfectly at all time points except one control rat at post-TBI day 4 only (4.2%). The TBI group had non-zero sound reflex scores (reflecting any non-normal behavior) (Figure 4B) on post-TBI day 1 (score 1: 8.4%; score 2: 45.4%), day 4 (score 1: 5.7%; score 2: 11.4%) and day 8 (score 1: 5.56%; score 2: 5.7%).

Balance scores were the next largest contribution to the total NSS-R scores. The TBI group had increased balance scores (more deficits) (Figure 4C) on post-TBI day 1 (score 1: 22.7%; score 2: 25.56%), with a return towards baseline on day 4 (score 1: 14.3%; score 2: 8.57%) and day 8 (score 1: 8.57%; score 2: 2.86%).

Auditory Brainstem Responses

A subset of male and female rats in the TBI (n=15) and Sham groups (n=16) underwent ABR testing at 10, 16, and 20 kHz on post-TBI day 1 and day 8 (Table 4). Within each frequency, a repeated measures ANOVA (PROC GLM) was performed, which assessed the effects of group (TBI intervention with sex collapsed since there were no differences), time, and the interaction of group with time (Figure 5). At 10 kHz, there was a trend towards significant increase in the hearing threshold at post-TBI day 1 due to effects of TBI (p = 0.056) and time (p = 0.083) with a significant interaction (p = 0.027). At 16 kHz, there was a hearing threshold shift such that TBI rats had a higher threshold (more deficits) (p < 0.05) with a significant effect of time (p < 0.05) and a significant interaction (p < 0.01). At 20 kHz, there was a significant increase in ABR thresholds due to TBI (p < 0.001) and time (p < 0.05) with a significant interaction (p < 0.01). There were not differences between TBI and sham groups for any frequency at post-TBI day 8.

Table 4.

Auditory Brainstem Responses.

	Female Control			Female TBI			Male Control			Male TBI			p-value
	n	Mean	SD	n	Mean	SD	n	Mean	SD	n	Mean	SD	TBI	Time	Interaction
10kHz
Day 1	8	26	6	7	39	7	8	32	4	8	37	7	0.056	0.083	0.027
Day 8	6	30	9	4	28	6	8	30	4	7	34	6
16kHz
Day 1	8	28	7	7	36	8	8	26	4	8	34	6	0.032	0.027	0.006
Day 8	6	28	5	4	26	5	8	27	4	7	28	4
20kHz
Day 1	8	28	4	7	40	6	8	27	3	8	35	7	0.0008	0.017	0.009
Day 8	6	28	5	4	28	6	8	28	3	7	32	4

Hearing thresholds were not different between the sexes as shown and therefore collapsed in Figure 5 for comparisons between the TBI and control groups using a repeated measures ANOVA (PROC GLM). Significant values p<0.05 denoted in bold.

Figure 5. Auditory Brainstem Responses.

Assessments for hearing thresholds using auditory brainstem responses quantified with electrophysiological responses to acoustic stimuli at (A) post-TBI day 1 and (B) day 8. Auditory brainstem responses were tested in rats anesthetized using ketamine and dexmedetomidine. Individual data points are presented with the median, a box plot for the interquartile range, and error bars for the 95% confidence interval. Statistical significance denoted by **p<0.01; ***p<0.0001 according to one-way ANOVA.

Discussion

This study demonstrated that the RID model, at the biomechanical parameters targeted, is characterized by a delayed return of righting reflex, weight loss, and global neurologic dysfunction that returns to baseline over a week. This study did not convincingly demonstrate sex differences. The deficits in neurologic function were most prominent in the sound reflex (acoustic startle responses) and balance (vestibular function), which in humans have been associated with impaired emotional responsivity (31, 32) and anxiety (33), respectively. Therefore, additional studies are urgently needed to understand the pathophysiological mechanisms of TBI so that therapeutic interventions can be developed to improve patient-oriented outcomes.

Loss of consciousness after TBI is associated with worse outcomes and therefore is a prognostic feature in humans (34–36). RORR is a surrogate for loss of consciousness when anesthesia is necessary in a TBI model and has been used to equate injury severity between different TBI mechanisms (19). Delayed RORR is also associated with worse outcomes (36), which would be expected in animal TBI models and was observed in the data presented here (Figure 3, 4, and GLM). If RORR is to be utilized as a translational measurement given that loss of consciousness has prognostic value in humans, then it is critically important to avoid confounds related to the type and duration of anesthesia. We advocate for a consensus regarding standardization of the methodological and statistical approach, which should include normalization of RORR to the duration of anesthesia and inclusion of anesthesia duration as a covariate in statistical analyses. In our studies, the normalized RORR, but not the raw RORR time, was significantly different between the TBI compared to Sham group. The duration of anesthesia was not associated with either the raw or normalized RORR. This raises the intriguing hypothesis that anesthetic depth requirements are altered after TBI, which warrants further investigation since, to our knowledge, no studies have tested this in humans (Supplemental Materials) (28, 37–78).

Acoustic startle responses are impaired after TBI in humans (31, 32), rats (79), and mice (80) making it an invaluable translational research method (31, 32, 79, 80). After TBI, people have a decrease in startle potentiation to unpleasant images (31, 32, 81). These impairments are important because acoustic startle responses are associated with emotional deficits after TBI that may contribute to poor relationships due to lack of social self-awareness, empathy, adaptability, and the inability to recognize social cues and emotional expression (82, 83). These deficits can contribute to emotional detachment, callousness, (84) and social isolation (85), but treatment might be possible since after TBI, patients have been successfully trained to recognize emotion via facial affect (86) and supported with cognitive-behavioral group therapy that addresses emotional impairments impeding social competence (87).

In this study, the RID model caused deficits in acoustic startle responses, such that the behaviors were blunted. This is opposite to the excessive acoustic startle responses that are quantified in fear behavioral phenotypes and are commonly assessed in the diagnosis of PTSD, which has led others to hypothesize that blunted acoustic startle responses might mask aspects of exaggerated responses and interfere with the diagnosis of PTSD after TBI (88). However, startle responses could be confounded by hearing loss, which has rarely been controlled for except in some high-quality audiometry studies in military service Veterans (89). In preclinical studies, auditory threshold assessment is generally lacking. Others have verified that the startle deficit quantified by the NSS-R sound reflex corresponds to an acoustic startle response (24), but auditory threshold data to rule out hearing impairment was not performed. A strength of our study is the confirmation that startle deficits cannot be attributed to hearing impairments alone. Although there were statistically significant increases in the ABR thresholds, the increase from 25–30dB to 35–40dB would not likely be sufficient to impair the acoustic startle behaviors quantified in the NSS-R in response to a 60–70dB stimulus (Figure 4B). This is not to dismiss auditory processing or hearing impairments after TBI as being irrelevant. In fact, hearing deficits after TBI has been known about for decades (90), is an important health care issue for athletes (91) and Veterans (92) and can be due to peripheral or central defects in the hearing system. While highly prevalent and well-known in the setting of temporal bone fracture, hearing impairment can also occur in over 50% of patients after TBI without a bony fracture (93). Our findings demonstrate 5–15dB temporary threshold shifts, which is remarkably similar to the 10–15dB hearing loss seen in some human TBI studies (93). This supports this TBI model as a potential tool for studying TBI-related hearing loss in humans.

Limitations

This study has several limitations in addition to other well-known limitations of preclinical TBI models (Supplemental Materials) (94–96). First, in these studies, quantification of the acoustic startle response and balance was part of the NSS-R battery of functional and cognitive testing, which included a sound reflex test that is scored on a coarse 3-point scale (97). Future studies would benefit from higher sensitivity analyses, such as acoustic startle responses / pre-pulse inhibition and rotarod. Second, the underlying pathophysiological mechanisms responsible for the deficits in this study require further interrogation. For example, we cannot attribute the startle reflex deficits to a precise pathophysiological mechanism. Specific brain regions were not tested for causality and the lack of imaging (in vivo or histological) also limits the ability to consider associations with candidate sites. Therefore, further studies are warranted to specifically test the contributions of limbic and central autonomic networks on these processes, such as the prefrontal cortex and amygdala. Our study did not evaluate the temporal bones, tympanic membranes, or other aspects of the hearing conduction system. However, damage to these structures is an unlikely explanation since prior studies using this TBI rat model did not identify any changes suggestive of temporal bone fractures (17, 98, 99) and the ABR returned to baseline within 1 week, which would be too fast for structural damage. Furthermore, monitoring wave IV of ABR verifies that the signal from the acoustic stimuli is reaching the lateral lemniscus or inferior colliculus (100, 101). This does not exclude the possibility that there was irritation to the Labyrinthine artery that contributed to this mild temporary threshold shift, which could also have contributed to the balance beam issues identified. Although a less severe injury using this same TBI model has demonstrated MRI changes in both the frontal cortex and amygdala (17), which further supports pursuit of these as candidate regions for the effects seen here.

Conclusions

Improving outcomes after TBI is critically important to military service members, Veterans, and civilian patients that suffer from decreased function, anxiety (3), and dementia (4, 102). This rat TBI model characterized by loss of consciousness and body weight highlights deficits in global neurologic function, including acoustic startle responses that have been demonstrated after TBI in humans and rodents, and supports the need for appropriate quantification of emotional responses that impact quality of life. A better understanding of the complex pathophysiological mechanisms of brain injury patients that contribute to these issues is essential to optimize rehabilitation and develop new treatments.

Abbreviations:

TBI

Traumatic brain injury

RID

rotational injury device

NSS-R

Revised Neurobehavioral Severity Scale

ABR

Auditory Brainstem Responses