Mild Hyperthermia Aggravates Glucose Metabolic Consequences in Repetitive Concussion

Traumatic brain injury (TBI) is one of the leading causes of mortality and disability around the world. Mild TBI (mTBI) makes up approximately 80% of reported cases and often results in transient psychological abnormalities and cognitive disruption. At-risk populations for mTBI include athletes and other active individuals who may sustain repetitive concussive injury during periods of exercise and exertion when core temperatures are elevated. Previous studies have emphasized the impact that increased brain temperature has on adverse neurological outcomes. A lack of diagnostic tools to assess concussive mTBI limits the ability to effectively identify the post-concussive period during which the brain is uniquely susceptible to damage upon sustaining additional injury. Studies have suggested that a temporal window of increased vulnerability that exists corresponds to a period of injury-induced depression of cerebral glucose metabolism. In the current study, we sought to evaluate the relationship between repetitive concussion, local cerebral glucose metabolism, and brain temperature using the Marmarou weight drop model to generate mTBI. Animals were injured three consecutive times over a period of 7 days while exposed to either normothermic or hyperthermic temperatures for 15 min prior to and 1 h post each injury. A 14C-2-deoxy-d-glucose (2DG) autoradiography was used to measure local cerebral metabolic rate of glucose (lCMRGlc) in 10 diverse brain regions across nine bregma levels 8 days after the initial insult. We found that repetitive mTBI significantly decreased glucose utilization bilaterally in several cortical areas, such as the cingulate, visual, motor, and retrosplenial cortices, as well as in subcortical areas, including the caudate putamen and striatum, compared to sham control animals. lCMRGlc was significant in both normothermic and hyperthermic repetitive mTBI animals relative to the sham group, but to a greater degree when exposed to hyperthermic conditions. Taken together, we report significant injury-induced glucose hypometabolism after repetitive concussion in the brain, and additionally highlight the importance of temperature management in the acute period after brain injury.


Introduction
Traumatic brain injury (TBI), once referred to as the "silent epidemic", is now well recognized as a critical public health and socioeconomic problem. In 2013, approximately 2.8 million TBI-related had at least one recorded fever while in the intensive care unit (ICU). The presence of fever in the acute phase after TBI has been associated with neurological severity, significantly worsened functional outcomes, increased intracranial pressure, lower Glasgow Coma Scores, and longer stays in the ICU [21,22]. Preclinical studies have also shown a similar relationship between acute elevated brain temperature and aggravated outcomes [23][24][25][26]. Truettner and colleagues [23] reported that elevated brain temperature after mTBI altered the polarization of resting, anti-inflammatory M2-like macrophages/activated microglia towards a more proinflammatory, hyperactive M1-like phenotype, which is associated with aggravation of neurological deficits and sustained functional consequences. Sakurai et al. [25] demonstrated that neuropathological damage was exacerbated when the brain was hyperthermic (39 • C) directly prior to mTBI (mimicking environmental influences that cause elevated temperature pre-injury) versus hyperthermic conditions after mTBI (mimicking posttraumatic pyrexia). Titus and colleagues [24] showed significant long-term cognitive deficits after mTBI under pre-and post-TBI hyperthermic conditions. Thus, in at-risk populations, hot environments and/or heat-retaining outerwear coupled with exertion or exercise, conditions that military service personnel and athletes are often exposed to, is of significant concern and an important area of study.
Hyperthermia-induced consequences after TBI are multifactorial and can include aggravated blood-brain barrier disruption, robust inflammatory responses, elevated cytokine release, enhanced generation of reactive oxidative species, as well as increased prevalence of excitotoxic events [26][27][28]. These consequences are attributed to a greater degree of contusion formation, diffuse axonal injury, tissue atrophy, susceptibility of cortical and hippocampal neurons to cell death, and, important to the scope of this study, glucose metabolic perturbations [25,[28][29][30].
Because there appears to be a lack of knowledge in the field of glucose metabolism with regard to multiple, successive mTBI occurring while brain temperature is elevated, and because this is a commonly occurring phenomenon in certain high-risk populations, we sought to explore this relationship. In the present study, we assessed the metabolic state of the repetitively injured brain with and without exposure to hyperthermia.

Physiology
Physiological variables were consistent throughout the duration of all three TBI surgeries and the 14 C-2-deoxy-d-glucose (2DG) surgery. At time of trauma, repetitive mTBI normothermic animals were maintained between 36.7 and 36.9 • C while hyperthermic animals were kept between 38.6 and 38.9 • C ( Table 1). There were no changes in respiratory status during induction of hyperthermia prior to TBI.

Qualitative Metabolic Consequences
The closed-head weight drop model utilized in this study generates a mild diffuse concussive brain injury in the absence of focal brain lesions, and, thus, a clinically relevant tool to study repetitive concussion [14]. When assessing lCMR Glc in this model, it is important to evaluate multiple bregma levels encompassing a large array of cortical and subcortical structures. We evaluated nine bregma levels encompassing 10 diverse regions of interest (ROIs). Autoradiographic images showed normal patterns of glucose utilization in sham animals (Figure 1, Left). lCMR Glc levels were symmetrical in cortical and subcortical regions, typical of the awake rat in a resting-state. Three mild concussions over the course of one week (Day 0, Day 3, and Day 7 post initial injury) under normothermic conditions produced patterns of discernable depression in lCMR Glc relative to sham control animals. In all nine bregma levels evaluated, glucose utilization was considerably diminished in both cortical and subcortical structures bilaterally (Figure 1, Center). We found that hyperthermia exposure at each mTBI showed an even greater degree of glucose hypometabolism with all nine bregma levels exhibiting this pattern (Figure 1, Right). In order to facilitate qualitative comparisons of lCMR Glc intensity among groups, pseudocolor images were stacked in mirrored orientations ( Figure 2). conditions produced patterns of discernable depression in lCMRGlc relative to sham control animals. In all nine bregma levels evaluated, glucose utilization was considerably diminished in both cortical and subcortical structures bilaterally (Figure 1, Center). We found that hyperthermia exposure at each mTBI showed an even greater degree of glucose hypometabolism with all nine bregma levels exhibiting this pattern (Figure 1, Right). In order to facilitate qualitative comparisons of lCMRGlc intensity among groups, pseudocolor images were stacked in mirrored orientations ( Figure 2). Quantitative autoradiographic images of local cerebral metabolic rate of glucose (lCMRGlc) displayed in pseudocolor. Each coronal level (with reference to bregma) is the average of n = 7 or 8 brains from the three experimental groups: sham (n = 7), normothermia traumatic brain injury (TBI; n = 8), and hyperthermia TBI (n = 8). After repetitive mild concussive TBI (3× over 7 days), there was

Quantitative Findings
Two-way ANOVA and ROI pixel-based statistics confirmed visual observations that triple-hit repetitive mTBI significantly reduced glucose utilization relative to sham controls 8 days post initial TBI. When coupled with hyperthermia, there was an even greater degree of significant lCMRGlc depression. We detected pronounced glucose utilization deficits between sham and repetitive mTBI normothermic and hyperthermia groups in seven or eight (respectively) of the nine bregma levels analyzed (Table 2). Table 2. Values represent local cerebral metabolic rate of glucose as mean ± standard error of the mean (µmol/100g/min; n = 7-8 per group) of each region of interest (ROI) across nine bregma levels. Two-way ANOVA analyses showed significant differences between sham and triple-hit mild TBI (mTBI) groups. Hyperthermia conferred a higher degree of significance in affected ROIs. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001, where indicated against the sham group. There were no significant differences between the two TBI groups.

Region
Sham TBI Normo TBI Hyper

Quantitative Findings
Two-way ANOVA and ROI pixel-based statistics confirmed visual observations that triple-hit repetitive mTBI significantly reduced glucose utilization relative to sham controls 8 days post initial TBI. When coupled with hyperthermia, there was an even greater degree of significant lCMR Glc depression. We detected pronounced glucose utilization deficits between sham and repetitive mTBI normothermic and hyperthermia groups in seven or eight (respectively) of the nine bregma levels analyzed (Table 2). Table 2. Values represent local cerebral metabolic rate of glucose as mean ± standard error of the mean (µmol/100g/min; n = 7-8 per group) of each region of interest (ROI) across nine bregma levels. Two-way ANOVA analyses showed significant differences between sham and triple-hit mild TBI (mTBI) groups. Hyperthermia conferred a higher degree of significance in affected ROIs. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001, where indicated against the sham group. There were no significant differences between the two TBI groups. Intergroup differences were further visualized using summary plots of the 2DG radiographic levels of various structures at select bregma levels ( Figure 3). After repetitive mTBI, regardless of temperature group, we observed lCMR Glc depression across all ROIs at all bregma levels. We found that certain bregma levels and ROIs were more significantly affected by repetitive injury than others. The cingulate cortex, retrosplenial cortex, primary and secondary motor cortices, as well as the cortical strip were some of these regions. The retrosplenial cortex in particular exhibited susceptibility to hyperthermia. Our quantitative data also showed that bregma level −1.8 encompassed the greatest number of ROIs that were significantly hypometabolic in glucose utilization relative to sham control animals. In the right cingulate cortex at −1.8 mm to bregma, there was a 36% decrease in 2DG levels in the repetitive mTBI normothermic group compared to sham animals, while hyperthermia reduced lCMR Glc by 47% relative to sham controls. Repetitive mTBI coupled with hyperthermia at this bregma level decreased lCMR Glc in the primary and secondary motor cortices 48% and 43% respectively. We did not observe significant decreases in glucose utilization between normothermic and hyperthermic repetitive injury groups although values maintained a greater degree of reduction with hyperthermia exposure.  Quantitative analysis of 2DG values confirmed significantly decreased glucose utilization in both repetitive mTBI groups relative to sham, though no significance was attained when comparing the two triple-hit groups. However, there was a marked difference in number of affected ROIs as well as in degree of significance when TBIs occurred under hyperthermic conditions. Nine ROIs at eight bregma levels had significantly decreased lCMR Glc values bilaterally: cingulate cortex (+0.2, 3). However, significant injury-induced depression in glucose metabolism was not observed in the hippocampus relative to sham controls.

Discussion
By using the Marmarou model of concussive injury and 2DG autoradiography, we found that three mild concussions over a 7 day span were associated with significantly depressed glucose metabolic activity relative to noninjured controls in several cortical and subcortical regions 8 days post initial TBI. Elevated core temperature at the time of traumatic insult resulted in a greater and more significant degree of metabolic depression relative to control animals. Abnormal hypometabolic alterations in glucose utilization were observed within several diverse, both functionally and spatially, regions. Deficient glucose metabolism occurred in primary and secondary motor, cingulate, retrosplenial, parietal association, and visual cortices, which are structures located medially beneath the sagittal suture and parietal bones (where weight drop impact occurs). These affected areas are responsible for higher-order cognitive processes, such as integrated motor control and goal-directed behavior, reward, spatial decision-making, and navigation. Depression in glucose metabolism was also observed in the caudate putamen and striatum, two subcortical structures that contribute to movement initiation, regulation, and certain types of learning. As we looked more posteriorly, we detected deficient glucose utilization in the medial temporal lobes, specifically within the ectorhinal and perirhinal cortices, areas involved in recognition memory, perception, fear conditioning, and spatial awareness. It would be important to do relevant behavioral tests assessing functional outcomes to determine correlation, if any, between affected structures and resulting behavior.
In the present study, lCMR Glc values appeared somewhat elevated in the sham group (n = 7) with the ROI heat maps corroborating this observation. In generating mild concussive TBI, all animals are fasted prior to any surgical procedures to maintain consistent baseline glucose levels, which can alter post-injury outcomes [31]. For this autoradiographic repetitive injury study, all animals are fasted 4 times (3× prior to each TBI or sham surgery, and 1× before 2DG procedure) over a span of 8 days, which may have resulted in hypoglycemia [32]. Furthermore, isoflurane anesthesia has been shown to reduce baseline glucose metabolism [33]. Previous studies have shown that under hypoglycemic conditions, there is an upregulation of glucose transporter-3 (GLUT3) as a cerebral adaptation [34]. Thus, when assessing glucose metabolism during the 2DG procedure (Day 8) greater cerebral expression of GLUT3 may have resulted in a larger degree of glucose uptake. However, given that all randomized experimental groups underwent identical conditions, this potential variability would have been consistent between all animal groups.
There is a great need for establishing evidence-based guidelines to determine when after sustaining mTBI it is safe to return to normal activity. Determination of the rate of glycolysis after injury has been suggested to be a valuable diagnostic tool for patients with TBI [10,35]. The temporal changes of injury-induced depression of glucose utilization represent a period in which the brain is particularly susceptible to further damage and the severity of glucose depression is associated with functional deficits and outcomes [4,15]. Several preclinical and clinical studies have shown worsened outcomes with repeat concussions [10,12,15]. However, the degree to which elevated temperature contributes to glucose metabolic deficits is less known. [ 18 F]fluorodeoxyglucose-positron emission tomography (FDG-PET) can be utilized to assess the real-time temporal patterns of glucose metabolism and correlation of functional and metabolic recovery [36]. Quantifying the degree of lCMR Glc may serve as an important biomarker to determine severity of injury, temporal course of greater vulnerability, as well as determination of treatment paradigms.
The effects of small elevations in brain temperature on cerebral metabolism have been previously documented [37][38][39]. In one study, Mickley and colleagues (1997) reported that an induced rise of 2 • C induced by hot/moist air caused a general increase in brain glucose metabolic rate [37]. In a human study, an acute exposure of whole body heat increased glucose and insulin levels in healthy individuals [38]. Regional changes in cerebral metabolism during systemic hyperthermia induced by a liquid-conditioned suit were also reported to increase in several brain regions (hypothalamus, thalamus, corpus callosum, cingulate gyrus and cerebellum) while decreasing in others (caudate, putamen, insula and posterior cingulum) using positron emission tomography [39]. Taken together, these studies emphasize the complex effects of elevated temperatures on regional patterns of glucose metabolism that could impact the long term consequences of TBI especially when present at the time of the insult.
In reference to potential therapeutic interventions, targeted temperature management and therapeutic hypothermia after moderate and severe TBI have been evaluated in multiple randomized controlled trials (RCTs) with inconsistent results [40,41]. In contrast, preclinical translational findings have been encouraging and have provided evidence that cooling the brain after TBI reduces secondary injury mechanisms that contribute to histopathological abnormalities and functional deficits [42]. Currently, there are limited data regarding the effects of brain cooling when core temperature at the time of a mild TBI is elevated. Interestingly, Walter and colleagues [43] have recently reported that selective brain cooling in student athletes in the acute post-concussive period restored an injury-induced reduction of cerebral blood flow (CBF). In that study, post-concussive brain cooling also resulted in the temporary relief of self-reported physical symptoms [43]. Our laboratory has reported in rodents that posttraumatic cooling reduces the cognitive deficits observed after hyperthermic mTBI [24]. Hyperthermia significantly aggravates TBI-induced deleterious events, including a prolonged and a greater degree of depression of glucose metabolism. Thus, targeted temperature strategies that address the potentially deleterious consequences of elevated brain temperatures may be important in the acute management of mild TBI and provide a clinically relevant approach to improving outcomes in this growing patient population.

Weight Drop Injury
All studies conducted were approved by the University of Miami Animal Care and Use Committee and by the NIH "Guide for the Care and Use of Laboratory Animals". In total, 23 male Sprague-Dawley rats were designated to one of three groups: Sham + normothermic (n = 3), Sham + hyperthermic (n = 4), repetitive mTBI + normothermia (n = 8), and repetitive mTBI + hyperthermia (n = 8). Animals were housed in their home cages for one week before any experimental procedures were conducted. Rats had access to food and water ad libitum and were exposed to a 12 h light/dark cycle.
To generate a mild concussive injury, we utilized the Marmarou weight drop model [44,45]. The injury device consists of a segmented brass weight free-falling through a Plexiglas tube. Fasted animals were anesthetized with 3% isoflurane in 30% O 2 /70% N 2 O and placed in a stereotaxic frame. The animal's head was shaved, and a midline incision was made allowing access to the periosteum covering the vertex of the skull. Once the periosteum was reflected, a stainless-steel disk (10 mm in diameter; 3 mm thick) was secured in the center of the intact skull at the cross-section of bregma and midline with cyanoacrylic adhesive. This disk served as an injury helmet, which protected the animal from skull fracture. After the disk was fully adhered to skull, the animal was removed from the stereotaxic table and maintained under anesthesia via nosecone. Rectal and temporalis muscle thermometers measured and maintained body and brain temperatures using self-adjusting feedback warming lamps. Animals were maintained at normothermic (36.7-37 • C) or hyperthermic (38.7-39 • C) conditions for 15 min prior to injury and for 1 h post injury. Temperature was manipulated using the self-adjusting feedback warming lamps. Desired temperature (37 or 39 • C) was set and warming lamps remained on until the set temperature was reached, at which point the lamps would automatically shut off. If temperature began to drop, the feedback lamps would turn on again. Body and head temperatures were measured and manipulated separately to ensure greater accuracy.
To induce mTBI, animals were removed from anesthesia and placed in the prone position on a foam pad underlying the weight drop device. The injury helmet affixed to the skull was centered directly below the Plexiglas tube where a 450 g weight was allowed to free fall from a height of 1.0 m to generate a mild concussive-like TBI. This height and weight combination results in a brain acceleration of 900 G and a brain compression gradient of 0.28 mm, generating a mild diffuse brain injury. Immediately after injury, the animal was returned to anesthesia and monitored for respiratory changes. Once respiration was normalized, the stainless steel disk was removed and the skull was inspected for fracture. If skull fracture was observed, the animal was discarded from the study. Body and brain temperatures were maintained at 37 or 39 • C for 1 h, at which point anesthesia was terminated. Once conscious, animals were administered a subcutaneous injection of 0.03 mg/kg buprenorphine for pain management and returned to their home cage. Normothermia and hyperthermia injury groups received three injuries: on Day 0, Day 3, and Day 7. The injury interval paradigm selected is clinically representative of repetitive concussion in sports-related injuries. In a prospective study of 16,624 high school and collegiate athletes, McCrea et al. [46] showed that most second or third repeat concussions (79.2%) occurred within 10 days of the initial injury.

14 C-2-Deoxy-d-Glucose (2DG) Surgical Methods
Eight days after the first mTBI, fasted animals were reanesthetized for insertion of femoral arterial and venous catheters, as well as a rectal thermometer to maintain normothermic temperatures throughout the duration of the 2DG experiment. Animals were wrapped in a loosely-fitting plaster body cast (Gypsona plaster of Paris), which was secured to a lead block. Anesthesia was terminated and animals were allowed 1 h to recover in a quiet and dark room.
Before the start of the experiment, a blank blood sample was obtained from the femoral artery for baseline radiation levels and baseline glucose concentration. At start, a pulse injection of 20 µCi of 14 C-labeled deoxy-d-glucose (Perkin Elmer #NEC495) dissolved in 0.2 cc isotonic saline was administered through the femoral vein. Twenty-four arterial blood samples were withdrawn at various intervals over a 45 min period as previously described [47]. After 45 min had elapsed, animals were euthanized via venous injection of Euthasol. Plasma aliquots from samples were assayed for 2DG using a liquid scintillation counter (Beckman) and glucose concentration was assessed via an automated glucose analyzer (YSI 2300). After euthanization, animals were decapitated. Whole brains were quickly removed and frozen over liquid nitrogen vapor. Frozen brains were cryosectioned subserially into 30 µm thick sections (Leica CM1900). A total of 18 bregma levels were collected ranging from 1.2 to −7.3 mm relative to bregma. Brain sections and [ 14 C] methylmethacrylate standards were exposed to BioMax MR high-resolution film for 10 days.

Image Processing
Autoradiographic films were digitized at 10 bit intensity precision by a charge-coupled device camera. The camera was interfaced to an advanced image acquisition/analysis system where images were captured at 70 µm/pixel in spatial resolution. [ 14 C] methylmethacrylate standards placed on the film were digitized in parallel to permit conversion of optimal density values to activity units of nano-curies per gram of equivalent tissue mass [48]. Images for each experimental group were then transferred to a Linux operating system computer for image registration and averaging of corresponding coronal sections from individual animals.

Region of Interest Analysis and Statistical Analysis
In order to assess the lCMR Glc of particular regions and structures, we selected nine coronal bregma levels (+0.7, +0.2, −0.8, −1.8, −2.8, −3.8, −4.8, −5.8, −7.3 mm relative to bregma) from the images of each animal that encompass the area of the brain anterior to, posterior to, and directly underneath the impact site where weight drop occurs (approximately 0.0 mm relative to bregma). The Marmarou model of closed head injury utilized in this study generates modest neuronal morphological changes in the supraventricular cortex, which is comprised of several of the ROIs used in this study: cingulate cortex (Cg), primary and secondary motor cortices (M1 and M2, respectively), the retrosplenial cortex (RS), the parietal association cortex (PtA), as well as the visual cortex (VM). We also chose to evaluate lCMR Glc in the caudate putamen (CPu), the striatum (STR), the hippocampus (CA), and cortical strip (Cor). Figure 4 illustrates three representative sections showing the 10 cortical and subcortical ROIs, which contribute to a wide array of functional processes. (Cor). Figure 4 illustrates three representative sections showing the 10 cortical and subcortical ROIs, which contribute to a wide array of functional processes. Nine digital brain atlas sections were prepared [49]. Sections at the same bregma level from each individual rat were mapped into the brain atlas section at the corresponding bregma level. The image registration procedure was based on an established disparity analysis algorithm whose theory and practical validation were described previously in detail [50,51]. Intergroup differences for lCMRGlc were evaluated using two-way ANOVA. Since the brain sections were mapped into the contours of the corresponding atlas sections, data collection was conducted by superimposing the atlas with the outlined ROI on top of the deformed brain section so that data within the defined ROI could be accurately acquired [52]. We observed no differences between the Sham normothermic and hyperthermic groups. Thus, the groups were combined into a single Sham group for analysis.  Acknowledgments: Ofelia Furones-Alonso taught several of the methodologies utilized in this study.

Conflicts of Interest:
The authors declare no conflicts of interest.

TBI
Traumatic brain injury 2DG 14 C-2-deoxy-D-glucose mTBI Mild traumatic brain injury lCMRGlc Local cerebral metabolic rate of glucose CBF Cerebral blood flow CNS Central nervous system Nine digital brain atlas sections were prepared [49]. Sections at the same bregma level from each individual rat were mapped into the brain atlas section at the corresponding bregma level. The image registration procedure was based on an established disparity analysis algorithm whose theory and practical validation were described previously in detail [50,51]. Intergroup differences for lCMR Glc were evaluated using two-way ANOVA. Since the brain sections were mapped into the contours of the corresponding atlas sections, data collection was conducted by superimposing the atlas with the outlined ROI on top of the deformed brain section so that data within the defined ROI could be accurately acquired [52]. We observed no differences between the Sham normothermic and hyperthermic groups. Thus, the groups were combined into a single Sham group for analysis.