Glibenclamide Treatment in Traumatic Brain Injury: Operation Brain Trauma Therapy

Abstract

Glibenclamide (GLY) is the sixth drug tested by the Operation Brain Trauma Therapy (OBTT) consortium based on substantial pre-clinical evidence of benefit in traumatic brain injury (TBI). Adult Sprague-Dawley rats underwent fluid percussion injury (FPI; n = 45), controlled cortical impact (CCI; n = 30), or penetrating ballistic-like brain injury (PBBI; n = 36). Efficacy of GLY treatment (10-μg/kg intraperitoneal loading dose at 10 min post-injury, followed by a continuous 7-day subcutaneous infusion [0.2 μg/h]) on motor, cognitive, neuropathological, and biomarker outcomes was assessed across models. GLY improved motor outcome versus vehicle in FPI (cylinder task, p < 0.05) and CCI (beam balance, p < 0.05; beam walk, p < 0.05). In FPI, GLY did not benefit any other outcome, whereas in CCI, it reduced 21-day lesion volume versus vehicle (p < 0.05). On Morris water maze testing in CCI, GLY worsened performance on hidden platform latency testing versus sham (p < 0.05), but not versus TBI vehicle. In PBBI, GLY did not improve any outcome. Blood levels of glial fibrillary acidic protein and ubiquitin carboxyl terminal hydrolase-1 at 24 h did not show significant treatment-induced changes. In summary, GLY showed the greatest benefit in CCI, with positive effects on motor and neuropathological outcomes. GLY is the second-highest–scoring agent overall tested by OBTT and the only drug to reduce lesion volume after CCI. Our findings suggest that leveraging the use of a TBI model-based phenotype to guide treatment (i.e., GLY in contusion) might represent a strategic choice to accelerate drug development in clinical trials and, ultimately, achieve precision medicine in TBI.

Introduction

Traumatic brain injury (TBI) is a major and pressing global health priority, afflicting >50 million persons worldwide each year and with wide-ranging consequences.^1,2 Despite progress in the diagnosis and management of patients,^1,3–5 to date no effective treatment has been shown to produce any clinically relevant benefit in long-term outcome. In the Operation Brain Trauma Therapy (OBTT) consortium, novel and promising acute therapies are tested across multiple pre-clinical TBI models using a screening approach and a panel of behavioral, histological, and biomarker outcomes to gauge treatment efficacy.^6–9 This novel and rigorous framework seeks to translate neuroprotection successes from the laboratory to the clinical setting by both selecting the best drugs to move forward to clinical trial and informing trial design by identifying therapeutic efficacy to targeted injury mechanisms toward achieving a precision medicine approach.

Glibenclamide (a.k.a., glyburide [GLY]), is the sixth drug selected for cross-model testing by the OBTT consortium. Long used to treat diabetes mellitus, GLY is a second-generation sulfonylurea that binds to the sulfonylurea receptor (SUR) subunit of adenosine triphosphate (ATP)-sensitive potassium channels (KIR6.2/KCNJ11).¹⁰ In the early 2000s, the promising role of GLY in central nervous system (CNS) injury was identified in seminal work by Simard and colleagues, reporting the discovery of a novel channel (sulfonylurea receptor 1/transient receptor potential melastatin 4; SUR1-TRPM4) in injured CNS cells that was susceptible to sulfonylurea inhibition.^11–13

Transcriptional upregulation of SUR1 and de novo formation of the SUR1-TRPM4 channel have been reported in multiple cell types of the neurovascular unit in several pre-clinical models of CNS injury, including stroke, subarachnoid hemorrhage, intracranial hemorrhage, encephalitis, and TBI, rendering it a compelling therapeutic target.^13–21 Emerging evidence indicates that GLY acts by inhibiting the channel activation that results in oncotic edema, which, in turn, leads to cell blebbing and oncotic cell death.^15,18 Although GLY does not cross an intact blood–brain barrier (BBB), its CNS entry after injury is facilitated by BBB disruption and decreased pH in injured/ischemic tissue from lactic acidosis.^22,23 It can also directly act on endothelial cells without BBB penetration.

The preponderance of evidence supporting the clinical effectiveness of GLY has been in stroke and has led to the initiation of an ongoing phase III study (NCT02864953).^13,23–31 Though less established, there is a growing body of pre-clinical and clinical research supporting the role of the SUR1-TRPM4 pathway and potential benefit of GLY in TBI.^{17,21,32–42} GLY treatment has improved measures of intracranial pressure, cerebral edema, BBB dysfunction, contusion expansion, hippocampal damage, and memory dysfunction in rodent models of controlled cortical impact (CCI), CCI with hypotension, and rat blast TBI.^{17,21,32,33,39,43,44} Whereas the putative benefit of GLY in many of these studies was by means of SUR1-TRPM4 inhibition, neuroprotective properties also include downregulation of c-Jun N-terminal kinase/c-jun apoptotic signaling, zona occludens-1 (a tight junction protein), and nucleotide-binding domain leucine-rich repeats family protein-3.^44,45

In human TBI, SUR1 expression has been detected after contusion in various cell types (neurons, astrocytes, endothelial cells, microglia, macrophages, and neutrophils) with different temporal patterns.⁴⁶ Cerebrospinal fluid (CSF) SUR1 levels were undetectable in uninjured patients, but exhibited a sharp rise after TBI, with trajectories that correlated with outcome and intracranial hypertension.⁴² Moreover, genetic variation in ATP-binding cassette transporter sub-family C member 8 (encoding SUR1) and TRPM4 may influence cerebral edema generation, intracranial hypertension, and outcome after TBI—revealing a potential new avenue for genotype-guided interventions.^{37,38,40,41,47} Small, randomized trials of oral GLY suggest a decreased contusion expansion rate in moderate-severe TBI and improved short-term outcome after diffuse axonal injury.^35,36 A pilot (n = 7 per group) phase II study indicated a trend toward reduced contusion volumes in TBI patients treated with intravenous GLY.³⁴ A larger phase II study is ongoing (ASTRAL, NCT03954041).

On the basis of these findings, and the fact that a therapy targeting cerebral edema had not been previously tested, the OBTT consortium designed this study to evaluate the efficacy of GLY to improve outcome metrics (i.e., sensorimotor and cognitive function, neuropathology, and blood biomarkers) across its three established pre-clinical rodent TBI models—fluid percussion injury (FPI), CCI, and penetrating ballistic-like brain injury (PBBI). GLY dosing was based on discussions with, and previously published research by, Simard and colleagues.²⁷

Methods

OBTT studies (including testing of GLY reported in this article) were executed under a standardized paradigm, including animals, models, outcome metrics, and blinding protocols.⁴⁸ General methodological details for this study are identical to those presented in earlier OBTT publications and are summarized below.^48,49 Nuances specific to GLY are emphasized and include pharmacokinetic testing, glucose-level determination, and measurement of cerebral edema as secondary outcomes.

Animals

Adult male Sprague-Dawley rats (270–320 g) were used for all experiments. Rats were housed individually under a 12-h light/dark cycle. All procedures were approved by each institution's individual institutional animal care and use committee and by the Animal Care and Use Review Office, U.S. Army Medical Research and Materiel Command. Experiments were conducted in compliance with the Animal Welfare Act and adhered to the principles stated in the Guide for the Care and Use of Laboratory Animals (National Research Council, Washington, D.C.) and other federal statutes and regulations relating to experiments involving animals.

Animal models

Fluid percussion injury model, University of Miami

As previously described, rats were anesthetized with 70% N₂O, 1–3% isoflurane, and 30% O₂ 24 h before injury and surgically prepared for parasagittal FPI.^48,49 A 4.8-mm craniotomy was performed 3.8 mm posterior to bregma and 2.5 mm lateral to midline. A plastic injury tube was placed over the exposed dura and affixed to the skull with adhesive and dental acrylic. The scalp was subsequently sutured closed, and rats were allowed to recover before returning to their cages. After fasting overnight, rats were reanesthetized with the aforementioned gas mixture, tracheally intubated, mechanically ventilated, and had catheters inserted into the tail artery (to monitor arterial blood pressure and blood gases) and the jugular vein (blood sampling for biomarker analysis). They were then subjected to the FPI, which entailed a pressure pulse of moderate (1.8–2.2 atm) intensity.⁵⁰ After TBI, rats were returned to their home cages with food and water available ad libitum. Sham rats underwent all procedures except FPI.

As in previous OBTT studies, FPI served as the sentinel model to assess effects of therapies on acute physiological parameters, including hemodynamics and blood gases. Arterial blood gas levels were obtained 15 min pre- and post-FPI. Brain temperature was measured indirectly by a thermistor probe placed in the right temporalis muscle. Core temperature was measured by rectal thermistor probe.

Controlled cortical impact model, University of Pittsburgh

Rats were anesthetized (4% isoflurane in 2:1 N₂O/O₂) and tracheally intubated with a 14-gauge angiocatheter. Anesthesia was maintained using 2% isoflurane in 2:1 N₂O/O₂. After intubation, rats were mechanically ventilated and placed on a thermal blanket to regulate body temperature (37°C) with the head placed in a stereotaxic frame. A parasagittal craniectomy (anteroposterior [AP] center of craniectomy, +4.0 mm; lateral, +2.8 mm from lambda) 8 mm in diameter was performed to expose the brain to allow access for the impactor tip of the CCI device (Pittsburgh Precision Instruments, Inc., Pittsburgh, PA). CCI was performed at a depth of 2.6 mm at 4 m/s.⁵¹ After CCI, the surgical site was closed by silk sutures, and rat recovery was monitored by measuring tail pinch and righting reflexes. Sham rats underwent all procedures except CCI.

Penetrating ballistic-like brain injury model, Walter Reed Army Institute of Research

Surgical procedures were performed using isoflurane anesthesia (3–5% for induction and 2% for maintenance by nose cone) with aseptic conditions and monitoring of physiological vital signs. PBBI was executed as previously reported.⁵² Anesthetized rats were placed on a thermal blanket to regulate body temperature (37°C), and the head was secured in the stereotaxic device to insert the PBBI probe. After a midline scalp incision, a right frontal cranial window (diameter = 4 mm) was created using a dental drill to expose the right frontal pole (+4.5 mm AP, +2 mm mediolateral to bregma). The PBBI probe was advanced through the cranial window into the right hemisphere to a depth of 1.2 cm from the brain surface. Once the probe was in place, the pulse generator was activated by a computer to release a pressure pulse calibrated to produce a rapid expansion of the water-filled elastic tubing that induced an elliptically shaped balloon (diameter = 0.633 mm, duration = 40 ms) to a volume equal to 10% of total brain volume. After deflation, the probe was manually retracted from the brain, and the cranial opening was sealed with sterile bone wax. The skin incision was closed with wound clips. Sham rats underwent all procedures except insertion of the PBBI probe.

Drug administration

GLY was purchased from Sigma-Aldrich (catalog number, G2539; Sigma-Aldrich, St. Louis, MO). A stock solution of 2.5 μg/μL of GLY was made weekly in 100% dimethyl sulfoxide (DMSO) and stored at room temperature (RT). In all models, a loading dose was administered intraperitoneally (i.p.), followed by a 7-day maintenance infusion using subcutaneous (s.c.) miniosmotic pumps (Alzet 2001; Alzet Corp, Cupertino, CA). Specifically, the loading dose was prepared by adding 4 μL of stock solution to 1 mL of sterile unbuffered normal saline (NS). In all models, the loading-dose solution (10 μg/kg) was given i.p. 10 min post-injury. The maintenance dose was a 7-day continuous s.c. infusion at 0.2 μg/h (using Alzet miniosmotic pumps). Pumps were loaded with 4 μL of 10-N Fixanal solution, 2.3 mL of sterile NS, and 200 μL of GLY stock solution.

Vehicle solutions were made with DMSO, NS, and Fixanal exactly as described above; however, GLY stock solution was omitted. Pumps were filled with a measured volume of solution (222–238 μL), with volume confirmed by the difference in mass before and after loading. Pumps were then primed overnight in sterile NS at 37°C to allow drug delivery immediately upon placement. Pumps were inserted s.c. immediately after the i.p. bolus. At each study site, drug was prepared and coded by persons other than those who performed the injury and/or outcome assessments. Table 1 summarizes experimental group sizes for each site and TBI model.

Table 1.

Summary of Experimental Group Sizes for Traumatic Brain Injury/Glibenclamide Study^a,b,c

Group	Sham	TBI-vehicle	TBI treated	N
FPI, Miami	15	15	15	45
CCI, Pittsburgh	10	10	10	30
PBBI, WRAIR	12	12	12	36

N indicates sample size.

^aSeparate sham rats, treated, and vehicle (n = 5 per group) were studied for serial monitoring of plasma levels of glibenclamide and serial assessment of blood glucose. See text for details.

^bSeparate CCI-injured rats treated with glibenclamide (n = 2) were studied to assess the ratio of CSF and plasma levels of glibenclamide at 24 h after injury to explore brain exposure. See text for details.

^cSeparate rats (n = 15) were used to assess the effect of glibenclamide treatment on brain edema in the CCI model at 24 h after injury. This included naïve, CCI plus vehicle treatment, and CCI plus glibenclamide treatment; n = 5 rats per group. See text for details.

FPI, fluid percussion injury; CCI, controlled cortical impact; PBBI, penetrating ballistic-like brain injury; WRAIR, Walter Reed Army Institute of Research; TBI, traumatic brain injury; CSF, cerebrospinal fluid.

Biomarker blood sample preparation

Details of the biomarker methods were previously described in earlier OBTT publications and are briefly summarized here.⁵³ Blood samples (0.7 mL) were collected at 1, 4, and 24 h post-injury and at the 21-day terminal end-point. Samples were collected from the jugular catheter in FPI and PBBI and from the tail artery using a heparinized syringe in CCI. Terminal samples were collected by cardiac puncture for all models. Immediately after collection, blood samples were transferred to 1.2-mL serum clotting tubes, stored at RT for 60 min (to allow clotting), and then placed on ice (to prevent protein degradation) until all samples were ready for centrifugation. Tubes were centrifuged at 5000g (RT) for 5 min. The supernatant was transferred into sterile 1.2-mL Eppendorf tubes, snap-frozen on dry ice, and stored at –80°C, pending analysis. Each sample was coded for study site, rat number, and sample collection time. Samples were shipped by FedEx priority overnight (on dry ice) to Banyan Biomarkers (Alachua, FL), where they were processed to detect ubiquitin carboxyl terminal hydrolase-1 (UCH-L1) and glial fibrillary acidic protein (GFAP).^49,54–56

Outcome metrics

Per the standardized paradigm of OBTT, the primary outcome metrics for this study are identical for all OBTT studies and are detailed in the introductory article of the 2016 special issue of the Journal of Neurotrauma.⁴⁸ These are summarized briefly for the FPI, CCI, and PBBI models: 1) Sensorimotor, 2) Cognitive, 3) Neuropathology, and 4) Biomarkers. Given the previously reported benefit of GLY on cerebral edema by means of putative inhibition of the Sur1-Trpm4 channel,^{13,15,33,39,44} percent brain water (%BW), as a secondary outcome, was assessed at 24 h in the CCI model.

Primary end-points

Sensorimotor testing

Fluid percussion injury model

Spontaneous forelimb use was measured using the forelimb asymmetry task.^49,57 Baseline measures were recorded immediately before FPI and again at 7 days post-injury. The number of times the rat placed either its right (ipsilateral to injury), left (contralateral to injury), or both forelimbs on the cylinder wall during rearing episodes was scored. Data were normalized using the following formula: Index of Asymmetry (IA) = (contra)/(ipsi + contra). The grid-walk task assessed fore- and hindlimb sensorimotor integration at 7 days post-injury. Rats were placed on a wire grid (25-mm square openings) for 5 min. The number of foot-faults per limb was recorded and expressed as a percentage of the total number of steps taken using that limb.

Controlled cortical impact model

Gross vestibulomotor function was assessed on a beam-balance task recording the amount of time the rat remained on an elevated, 1.5-cm-wide wooden beam (maximum = 60 sec). Rats were trained, and baseline performance was assessed the day before CCI. Fine vestibulomotor function and coordination were assessed using a modified beam-walking task that used aversive stimuli (i.e., bright light/loud noise) to motivate rats to traverse the beam and reach a darkened goal box.⁵⁸ Latency to traverse the beam was recorded. Rats were given three trials per day with a 30-sec intertrial interval (ITI) at 1 day before CCI (pre-injury baseline) and daily for 5 days post-CCI. The primary outcome measure for this task was mean latency (three trials) to traverse the beam.

Penetrating ballistic-like brain injury model

Neurological deficits were evaluated at 30 min and 1, 3, 7, 14, and 21 days post-injury using a modified battery of tests.⁵⁹ Neurological scores were based on a 12-point sliding scale ranging from 0 (normal) to 12 (severely impaired). These were recorded for 1) contralateral forelimb flexion during tail suspension, 2) shoulder adduction (body upward curling behavior) during tail suspension, 3) open-field circling behavior, and 4) impaired resistance to lateral push (maximum score for each component = 3). Motor coordination and balance were evaluated using a fixed-speed rotarod task, with the outcome measure for this task being mean latency (two trials) to fall during each successive speed increment (10, 15, and 20 rpm).^49,52

Cognitive testing

The Morris water maze (MWM) was used for cognitive testing at each site.^39,40 All trials were digitally recorded for computer-software–assisted analysis.

Fluid percussion injury model

Spatial learning performance was assessed from 13 to 16 d post-injury consisting of four trials per day with a 60-sec duration, 10-sec reinforcement, and 4-min ITI. Primary outcome metrics consisted of latency to locate the hidden platform and swim distance. Rats were tested for retention of the hidden platform location in a probe (missing platform) trial at 17 days post-injury.

Working memory was evaluated on post-injury days 20 and 21. Rats were given 60 sec to find a submerged (non-cued) platform. If the rat failed to locate the platform within 60 sec, it was placed on the platform for 10 sec. Five sec after trial 1 (Location) for the same rat, a second identical trial (Match) was conducted. Rats were placed under a heat lamp for 4 min between each paired trial. After running the group of rats as above, the platform was moved to the next location of the maze and the procedure was repeated with this location. Five paired trials were given for each rat on each testing day.

Controlled cortical impact model

Spatial learning performance was assessed from 14 to 18 days post-injury. Rats were given four trials per day with a 60-sec duration, 10-sec reinforcement, and 4-min ITI. Latency to locate the hidden platform was the primary outcome measure. Rats were tested for retention of the hidden platform location in a probe (missing platform) trial at 21 days post-injury. Rats were then tested on a visible platform task for 2 additional days (days 19–20) where the platform was raised 2 cm above the water's surface. The visible platform task was used as a control procedure to determine the contributions of non-spatial factors (sensorimotor performance, motivation, and visual acuity) on MWM performance.

Penetrating ballistic-like brain injury model

Spatial learning performance was assessed from 13 to 17 days post-TBI. Rats were given four trials per day with a 90-sec duration, 10-sec reinforcement, and 30-min ITI. Primary outcome metrics consisted of latency to locate the hidden platform and thigmotaxic (wall-hugging) behavior. Rats were tested for retention of the hidden platform location in a probe (missing platform) trial at 19 days post-injury.

Histopathological assessment

Rats were terminally anesthetized and transcardially perfused with 4% (FPI and PBBI) or 10% (CCI) phosphate-buffered paraformaldehyde at 21 days post-injury using previously established protocols at each site. Brains were processed for paraffin embedding/slicing (FPI and CCI) or frozen sectioning (PBBI) as previously described.^48,49 Serial 1-mm coronally sliced serial sections were processed with hematoxylin and eosin for quantitative volumetric analysis.^48,49 Lesion volume (mm³) was calculated by measuring the area of the lesion (mm²) and multiplying the sum of the lesioned areas obtained from each section by the intersection distance. Ipsi- and contralateral hemispheric tissue volume (CCI and PBBI) or cortical tissue volume (FPI) were quantified using the same approach, again using established protocols at each site. Both lesion- and tissue-volume loss were expressed as a percent of the contralateral (non-injured) hemisphere (CCI and PBBI) or as a percent of the contralateral cortex (FPI), in accordance with the established protocol given the small focal lesions in FPI.

Biomarker Assessment

Blood levels of UCH-L1 (neuronal cell body damage marker) and GFAP (astroglial cytoskeletal protein) were measured at 4 and 24 h post-injury.^48,53 Primary outcome metrics consisted of 1) evaluating effects of GLY on blood biomarker levels at 24 h post-injury and 2) effect of GLY on the difference between 4- and 24-h (delta 24–4 h) levels. Rat UCH-L1 and GFAP levels were assayed with sandwich enzyme-linked immunosorbent assay, as previously reported.^48,53–56

Operation Brain Trauma Therapy Outcome Scoring Matrix

As with all OBTT studies, primary outcome metrics were summarized in an overall outcome scoring matrix—the approach to this has been previously detailed.⁴⁸ This was constructed to rank the therapeutic efficacy of individual drugs across OBTT studies. Each therapy tested can generate a maximum of 22 points at each center (with cognitive outcome given the highest weight) and a maximum total score of 66 points across all three models (FPI, CCI, and PBBI).

Secondary end-points

Drug-level and glucose measurements

Two additional studies were performed. For these studies, separate rats from those in the OBTT outcome studies were used to ensure that additional blood sampling or other procedures did not affect any scored outcome. These studies were carried out at the Pittsburgh site in the CCI model and/or shams using the same protocols, including GLY and vehicle dosing, as in the primary experiments described earlier. First, GLY levels were determined in plasma (n = 5/group) in serial samples obtained at 1, 2, 4, 20, and 24 h after the bolus in sham rats administered either GLY or vehicle. GLY plasma levels were quantified by an ultra-performance liquid chromatography (UPLC)/mass spectrometer with tandem mass spectrometry (MS/MS). Given the potential for GLY to cause hypoglycemia, concurrent blood glucose levels were also assessed in these rats using a blood gas analyzer (Model ABL-90; Radiometer America, Westlake, OH). Second, in a second small series (n = 2), we explored the GLY level in CSF and plasma after CCI. For these experiments, plasma (from the jugular vein) and CSF (obtained by cisternal puncture) were collected at 24 h after CCI. The UPLC-MS/MS method used to quantify GLY has been previously described^39,60 and involved liquid-liquid extraction and detection with a triple quadrupole mass spectrometer.

Cerebral edema assessment

We also conducted separate studies to test the purported effect of GLY on cerebral edema. The CCI model was selected post hoc for these experiments given the more pronounced treatment effect that we observed in this model. Percent brain water (%BW) was quantified in three groups—naïve, CCI plus vehicle treatment, and CCI plus GLY treatment (n = 5 per group)—using the gold-standard wet/dry weight technique.⁴³ Rats were decapitated at 24 h after injury (<5% isoflurane and 50/50 gas mixture of nitrous oxide and oxygen). Brains were bisected into hemispheres, which were immediately weighed and recorded as wet weights. Hemispheres were subsequently dehydrated for 72 h in an oven (110°C) and reweighed for dry weights, again using an established protocol. %BW was determined by subtracting the dry from the wet weight, dividing this number by the wet weight, and multiplying by 100.

Statistical analysis

All parameters (functional, histological, and biomarker) were assessed for normality, and data are expressed as mean (standard error of the mean) or median (interquartile range), as appropriate. Physiological data, blood glucose, lesion and tissue volumes, and probe trial were analyzed using a one-way analysis of variance (ANOVA) or mixed-model regression. One-way ANOVA or repeated-measures ANOVA was used to analyze motor tasks, as appropriate, depending on the specifics of the data collection. Two-way repeated-measures ANOVA was also used to analyze serial data measurements for the hidden platform and working memory tasks. Post hoc analysis, when appropriate, used the Student-Newman-Keuls (SNK) test or Tukey's test.

Based on exploratory analysis, biomarker data were not normally distributed. Delta 24–4 h biomarker levels were calculated as the difference between 24- and 4-h levels. This measure of dynamic change in brain-injury markers identifies increases (positive delta) and decreases (negative delta) of levels in this 20-h epoch. Comparison of biomarker levels across groups in each TBI model was performed using the Mann-Whitney U test (in the case of two groups) and the Kruskal-Wallis test (if three groups), followed by post hoc comparisons while applying Dunn's test with a Benjamini and Hochberg adjustment. All hypothesis testing was two-tailed, and a p value <0.05 was considered significant. Statistical analysis was performed using SAS (SAS version 9.4; SAS Institute Inc., Cary, NC) and R software (http://www.r-project.org; version 3.5.1) in RStudio (http://www.rstudio.com; version 1.1.456).

Results

Physiological parameters

Blood pH, P_aO₂, P_aCO₂, mean arterial blood pressure (MAP), and brain and body temperature were recorded in the FPI model, as previously described (Table 2). Before injury, all parameters were within the normal physiological range. After injury, the GLY-treated group showed a modest, albeit significant, reduction in MAP versus both sham and vehicle-treated groups (p < 0.05), although values remained within normal range. Note that the 15-min MAP measurement was made at 5 min after administration of the i.p. loading dose of either GLY or vehicle.

Table 2.

Effect of Glibenclamide on Fluid Percussion Injury: Physiology

	Sham	TBI-vehicle	TBI-10 ug/kg	ANOVA	SNK (p value)
Pre-TBI
pH	7.43 ± 0.01	7.42 ± 0.01	7.41 ± 0.01	0.376
pO2 (mm Hg)	161.2 ± 6.82	160.2 ± 6.36	155.0 ± 4.94	0.744
pCO2 (mm Hg)	41.39 ± 0.77	40.1 ± 0.70	41.89 ± 0.87	0.260
MAP (mm Hg)	119.11 ± 2.07	124.41 ± 3.34	115.58 ± 1.85	0.054	2 vs. 3
Brain temp, °C	36.6 ± 0.02	36.6 ± 0.02	36.6 ± 0.02	1.000
Body temp, °C	36.5 ± 0.02	36.6 ± 0.03	36.6 ± 0.03	0.016	2, 3 vs. 1
Post-TBI
pH	7.44 ± 0.01	7.44 ± 0.01	7.43 ± 0.01	0.718
pO2 (mm Hg)	157.3 ± 6.05	158.0 ± 6.18	149.9 ± 4.77	0.543
pCO2 (mm Hg)	40.2 ± 0.71	38.05 ± 0.73	39.59 ± 0.77	0.117
MAP (mm Hg)	121.76 ± 1.43	121.14 ± 2.43	112.63 ± 1.62	0.002	1, 2 vs. 3
Brain temp, °C	36.6 ± 0.02	36.6 ± 0.02	36.6 ± 0.02	1.000
Body temp, °C	36.6 ± 0.02	36.6 ± 0.02	36.6 ± 0.02	1.000

TBI, traumatic brain injury; ANOVA, analysis of variance; SNK, Student-Newman-Keuls; MAP, mean arterial blood pressure, Temp, temperature.

Sensorimotor parameters

Fluid percussion injury model

At baseline, there was no difference in rat performance on the cylinder task for spontaneous forelimb use. One week after injury, the FPI-vehicle group showed a worsening versus both sham (p < 0.05) and GLY (p < 0.05; Fig. 1A). This indicated a recovery of GLY- versus vehicle-treated rats, generating the full points (+2) for GLY treatment on this task in the OBTT scoring matrix (Table 3).

FIG. 1.

Sensorimotor outcome. Fluid percussion injury (FPI) model (A,B): Bar graphs show the results of (A) cylinder task and (B) the gridwalk task. Controlled cortical impact (CCI) model (C,D): Line graphs show the results of the beam balance task, that is, the total time each rat remained on the elevated beam (C) and the beam walk task, that is, mean time taken to traverse the beam (D). Penetrating ballistic-like brain injury (PBBI) model (E–G): graphs showing results from the neuroscore evaluations (E), rotarod task (F), and mean-motor score (G). Glibenclamide (GLY) showed significant benefit versus TBI vehicle on the cylinder task after FPI and both beam balance and beam walking after CCI (all p < 0.05). Please see text for details and interpretation of the findings. Data represent group means ± standard error of the mean; *p < 0.05 versus sham; **p < 0.05 versus TBI-vehicle. rpm, revolutions per minute; TBI, traumatic brain injury; Veh, vehicle; sec, seconds.

Table 3.

Scoring Matrix for Assessment of Therapeutic Efficacy across Models in Operation Brain Trauma Therapy

Site	Neuro exam	Motor	Cognitive	Neuropathology	Serum biomarker	Model and overall total
Miami	None	Cylinder (2) Gridwalk (2)	Hidden platform latency (2) Hidden platform path length (2) MWM probe (2) Working memory latency (2) Working memory path length (2)	Lesion volume (2) Cortical volume (2)	GFAP: 24 h (1) 4–24 h Δ (1) UCH-L1: 24 h (1) 4–24 h Δ (1)
Miami total	N/A	4	10	4	4
Miami
Dose 1		+2, 0	0, 0, 0, 0	0, 0	0,0 0,0	+2
Pittsburgh	None	Beam balance (2) Beam walking (2)	Hidden platform latency (5) MWM probe (5)	Lesion volume (2) Hemispheric volume (2)	GFAP: 24 h (1) 4–24 h Δ (1) UCH-L1: 24 h (1) 4–24 h Δ (1)
Pittsburgh total	N/A	4	10	4	4
Pittsburgh
Dose 1		+2, +2	–2.5, 0	+2, 0	0,0 0,0	+3.5
WRAIR	Neuroscore	Rotarod (3)	Hidden platform latency (5) MWM probe (3) Thigmotaxis (2)	Lesion volume (2) Hemispheric volume (2)	GFAP: 24 h (1) 4–24 h Δ (1) UCH-L1: 24 h (1) 4–24 h Δ (1)
WRAIR total	1	3	10	4	4
WRAIR
Dose 1	0	0	0, 0, 0	0, 0, 0	0,0 0,0	0
Grand total
Dose 1	0	+6	−2.5	+2	0	+5.5

Drug: glibenclamide. Dose 1 = loading bolus (10 μg/kg) intraperitoneal at 10 min post-injury, followed by a 7-day continuous subcutaneous infusion at 0.2 μg/h (1-μL/h Alzet minipump).

( ) = point value for each outcome within each model.

MWM, Morris water maze, GFAP, glial fibrillary acidic protein, UCH-L1, ubiquitin C-terminal hydrolase-L1; N/A, not applicable; WRAIR, Walter Reed Army Institute of Research.

Figure 1B shows results of the grid-walk task, assessing fore- and hindlimb foot-faults as a percent of total steps for each limb. One-way ANOVAs for both ipsi- and contralateral fore- and hindlimb performance were not significant for differences between groups, demonstrating a limited injury effect and no treatment effect on this task. Thus, in FPI, 0 points were achieved for this task in the OBTT scoring matrix (Table 3).

Controlled cortical impact model

Figure 1C,D shows rat performance on beam-balance and beam-walking tasks after CCI. On beam-balance testing, two-way repeated-measures ANOVA showed a main group effect for latencies over 5 days (p < 0.05; Fig. 1C). Post hoc tests revealed that each group was different from the others (p < 0.05). The GLY group showed motor benefit over vehicle with improved latency (time spent on the beam without falling) after CCI (p < 0.05), thus scoring full points (+2) for this parameter in the OBTT scoring matrix (Table 3). Whereas initial latencies were reduced in both CCI groups (GLY and vehicle), by day 5 the GLY-treated group latencies approached sham (57.47 ± 1.93 and 60 sec, respectively), whereas vehicle group latency remained reduced (40.13 ± 8.05 sec).

Similarly, two-way repeated-measure ANOVA revealed a main group effect for beam walking over 5 days (p < 0.05; Fig. 1D). Post hoc tests again showed that each group was different from the others, with motor benefit indicated by reduced latency to traverse the beam in the GLY group versus vehicle after CCI (p < 0.05), thus generating full points (+2) in the OBTT scoring matrix for beam walking in favor of GLY therapy (Table 3). Neither injured group, however, returned to sham performance level over the 5-day testing period.

Penetrating ballistic-like brain injury model

Figure 1E–G shows neuroscore and rotarod performance after PBBI. Neuroscore assessments revealed abnormalities in all injured groups versus sham (p < 0.05) that persisted throughout the 7 days of testing (Fig. 1E). Post-injury treatment with GLY failed to provide benefit.

Rotarod performance was not different between groups pre-PBBI. At both 7 and 10 days post-PBBI, injured groups (vehicle and GLY) showed motor impairment with reduced latency to fall at all three speeds (10, 15, and 20 rpm, p < 0.05; Fig. 1F). As with the neuroscore assessment, post-injury treatment with GLY did not improve rotarod performance. Mean motor score, the parameter that is used to generate points on this task in our OBTT scoring matrix, showed reduced latency to fall in both injured groups (vehicle and GLY) versus sham (one-way ANOVA, p < 0.05; Fig. 1G). Thus, in PBBI, GLY treatment achieved no points for effects on neurofunctional or sensorimotor tasks in the OBTT scoring matrix (Table 3).

Cognitive testing

Fluid percussion injury model

FPI did not produce a significant increase in mean latency to finding the hidden platform versus sham, again suggesting a limited injury effect on this task in FPI (Fig. 2A). Consequently, there were no effects of GLY treatment versus either sham or vehicle, and GLY scored 0 points in the OBTT scoring matrix. Similarly, cognitive assessment on mean path length in the hidden platform task revealed the lack of an injury effect, limiting the possibility to assess a therapeutic effect of GLY on this task in FPI (Fig. 2B). No effect was identified for the hidden platform mean path length for GLY, which scored 0 points in the OBTT scoring matrix (Table 3).

FIG. 2.

Cognitive outcome. Fluid percussion injury (FPI) model (A–D): Graphs show spatial learning performance in the Morris water maze (MWM) task based on (A) latency and (B) path length to locate the hidden platform over 4 days of MWM testing. Working memory performance is represented by graphs showing the difference in (C) mean latency and (D) mean distance taken to reach the hidden platform. Controlled cortical impact (CCI) model (E): line graph showing the (E) latency to the hidden platform over 5 days of MWM testing. Penetrating ballistic-like brain injury (PBBI) model (F,G): graphs showing (F) mean latency to the hidden platform and (G) percent time spent circling the outer perimeter of the maze (thigmotaxic response) over 5 days of MWM testing. The CCI-gibenclamide (GLY) group had a worsened MWM performance versus sham (p = 0.001), whereas the CCI-vehicle group was not different from sham (p = 0.062), showing only a trend. Thus, GLY showed an intermediate worsening of MWM performance based on OBTT criteria generating negative points (−2.5) for this task in CCI in the scoring matrix. Pooled comparisons (H,I): Graphs show (H) the mean overall spatial learning performance (latency to locate the hidden platform) and (I) percent time searching the target zone during the probe (missing platform) trial. No treatment effects were observed on these pooled analysis parameters. Please see text for details and interpretation of the findings. Data represent group means – standard error of the mean; *p < 0.05 versus sham. OBTT, Operation Brain Trauma Therapy.

On the working memory task used to assess short-term recall memory, two-way repeated-measures ANOVA for working memory latency was not significant for group (p = 0.126), but was significant for trial (p < 0.001), given that rats in all groups located the platform faster on the second of the paired trials (Fig. 2C). There was no interaction effect (p = 0.834). GLY-treated rats had a trend toward improved (shorter) latency to locating the platform versus FPI-vehicle; however, this was not significant and yielded 0 points. The working memory task, in terms of path length traveled to find the platform, yielded similar results (Fig. 2D). Two-way repeated-measures ANOVA was not significant for group (p = 0.232), but was significant for the paired trial (p < 0.001), given that rats had shorter paths before they located the platform on the second trial. There was no interaction effect (p = 0.60); however, once again, the GLY-treated group showed only a trend toward a smaller distance traveled to locate the platform compared with FPI-vehicle. The FPI-vehicle group showed high variability. Again, GLY achieved 0 points in the OBTT scoring matrix for this task. Overall, in FPI, GLY scored 0 points for effects on cognitive outcome in the OBTT scoring matrix (Table 3).

Controlled cortical impact model

There was a significant group main effect by two-way repeated-measures ANOVA in the spatial memory performance test that measured daily swim latencies to find a hidden platform in the MWM (p = 0.001; Fig. 2E). Tukey's post hoc tests demonstrated that there was no difference in the primary outcome for this task, that is, between CCI-GLY versus CCI-vehicle groups (p = 0.23). However, the CCI-GLY group had a worsened MWM performance versus sham (p = 0.001), whereas the CCI-vehicle group was not different from sham (p = 0.062), showing only a trend. Thus, GLY showed an intermediate worsening of MWM performance, based on OBTT criteria, and this result generated negative points (−2.5) for GLY for this task in the scoring matrix. Overall, in CCI, GLY scored a total of −2.5 points in the OBTT scoring matrix for assessment of effect on cognitive outcome (Table 3).

Of note, 3 of the 10 rats in the CCI-vehicle group had significant functional impairment and could not swim (or be tested in the MWM), whereas only 1 rat in the CCI-GLY group could not swim, possibly affecting the results of the assessment of the treatment effect.

Penetrating ballistic-like brain injury model

Two-way repeated-measures ANOVA for latency to locate the hidden platform demonstrated a main group effect (p < 0.05; Fig. 2F). Post hoc analyses demonstrated no GLY treatment effect. GLY scored 0 points for MWM platform acquisition latency in PBBI.

Similarly, two-way repeated-measures ANOVA for thigmotaxic behavior yielded a main group effect (p < 0.05; Fig. 2G), with post hoc analysis showing no treatment effect with GLY and that all injured rats spent more time circling the maze periphery versus sham (p < 0.05). GLY scored 0 points for MWM thigmotaxis in PBBI. Overall, in PBBI, GLY scored a total of 0 points in the OBTT scoring matrix for effects on cognitive outcome (Table 3).

Pooled analysis of therapeutic effects across Operation Brain Trauma Therapy

Cognitive outcomes

For ease of comparison, a pooled analysis of average latency to find the hidden platform and probe trial results are presented across all three models in Figure 2H and 2I, respectively. In FPI, there was no difference in MWM average latency in any of the groups (sham, FPI-vehicle, and FPI-GLY), confirming the lack of an injury effect. However, in both CCI and PBBI, injured rats show increased average latency versus sham, but no benefit from GLY. In CCI, GLY-treated rats had worse MWM latency performance versus sham, consistent with the previously defined result of negative points (−2.5) in the OBTT scoring matrix; but using average latency, vehicle-treated CCI rats also differed from sham. Again, 3 CCI-vehicle rats were unable to be tested/swim versus 1 CCI-GLY rat because of motor deficits. However, mean latency is not a scored parameter in the OBTT scoring matrix, to avoid duplicating points from a second analysis of MWM latency outcomes.

In FPI and CCI, surprisingly, probe trial performance did not differ between groups, again, limiting the assessment of a potential therapeutic opportunity in either of these models for this outcome. However, in PBBI, both groups of injured rats (PBBI-vehicle and PBBI-GLY) spent less time in the target zone versus sham, but there was no treatment effect. Thus, the results of the probe trial did not generate any points in any of the models for the OBTT scoring matrix.

Histopathological outcomes

Cross-model comparisons of lesion volume and cortical (FPI) or hemispheric (CCI and PBBI) tissue loss are shown in Figure 3A,B.

FIG. 3.

Histopathology. Graphs show cross-model comparisons of (A) lesion volume and (B) tissue loss. In the CCI model, treatment with glibenclamide significantly reduced lesion volume versus TBI vehicle (p < 0.05), which generated a full +2 positive points in the OBTT scoring matrix. No other treatment effects on histology were observed. Please see text for details and interpretation of the findings. Data represent group means – standard error of the mean; *p < 0.05 versus sham; **p < 0.05 versus TBI + vehicle. CCI, controlled cortical impact; FPI, fluid percussion injury; GLY, glibenclamide; OBTT, Operation Brain Trauma Therapy; PBBI, penetrating ballistic-like brain injury; TBI, traumatic brain injury.

Fluid percussion injury model

There was no difference in lesion volume between FPI-vehicle–treated rats versus those treated with GLY (Fig. 3A; p = 0.97). Similarly, for cortical tissue loss, one-way ANOVA yielded a main group effect (Fig. 3B; p < 0.05). Post hoc analysis indicated that both injured groups (FPI-vehicle and FPI-GLY) showed more cortical tissue loss versus uninjured sham (p < 0.05). There was no benefit with GLY on either lesion volume or tissue loss after FPI. Overall, in FPI, GLY scored 0 points for effect on neuropathological outcome in the OBTT scoring matrix (Table 3).

Controlled cortical impact model

GLY treatment significantly reduced lesion volume versus vehicle after CCI (p = 0.039), displaying a 30% reduction. This generated a full +2 points for this outcome in the OBTT scoring matrix. Both vehicle and treated groups showed a significant increase in hemispheric tissue loss after CCI (Fig. 3B; p < 0.001). However, there was no significant difference between GLY and vehicle treatment, and, consequently, no points were awarded, despite a trend toward benefit. Overall, in CCI, GLY scored a total of +2 points in the OBTT scoring matrix for benefit on neuropathological outcome (Table 3).

Penetrating ballistic-like brain injury model

There was no difference in lesion volume between PBBI-vehicle– versus GLY-treated rats (p = 0.86; Fig. 3A), yielding 0 points on the OBTT scoring matrix. Both vehicle and treated groups showed a significant increase in hemispheric tissue loss after PBBI (one-way ANOVA main group effect for hemispheric tissue loss, Fig. 3B; p < 0.001). However, there was no difference between GLY and vehicle treatment; thus, no points were awarded. Overall, in PBBI, GLY scored 0 points for effect on neuropathological outcome in the OBTT scoring matrix (Table 3).

Biomarker assessments

To assess the effect of GLY on biomarker concentrations, we collected and analyzed blood samples from all 111 rats in the primary outcome matrix (Table 1). Biomarker data by model, group, and time point are presented in Figures 4 and 5 and Supplementary Figure S1.

FIG. 4.

Box plots illustrating circulating glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) levels at 4 and 24 h post-injury. GFAP and UCH-L1 concentrations at 4 and 24 h post-injury in FPI (A), CCI (B), and PBBI (C). The black horizontal line in each box represents the median, with the boxes representing interquartile range. Whiskers above and below the box indicate the 90th and 10th percentiles. Each individual value is plotted as a dot superimposed on the graph. Increases in GFAP after injury were observed across models, and increases in UCH-L1 were observed after FPI and PBBI. However, no treatment effects were noted at 24 h after injury—the pre-determined time point to assess the effect of treatment, and thus no points were awarded. Please see text for details. *p < 0.05, **p < 0.01, or ***p < 0.001 versus sham; ^#indicates p < 0.05 for vehicle versus gibenclamide (GLY). Comparison groups (vs. sham or vs. vehicle) are indicated by the horizontal lines outside the box plots. CCI, controlled cortical impact; FPI, fluid percussion injury; PBBI, penetrating ballistic-like brain injury.

FIG. 5.

Box plots illustrating delta (24–4 h) circulating glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) levels on the left and right panels, respectively. Delta 24–4 h GFAP and UCH-L1 levels in FPI (A), CCI (B), and PBBI (C). The black horizontal line in each box represents the median, with the boxes representing interquartile range. Whiskers above and below the box indicate the 90th and 10th percentiles. Each individual value is plotted as a dot superimposed on the graph. The red dotted lines on each graph are at a delta level of 0 ng/mL between 4 and 24 h. There were no significant differences between vehicle- and glibenclamide (GLY)-treated groups, indicating no treatment effect on this outcome across models, and thus no points were awarded. Please see text for details. CCI, controlled cortical impact; FPI, fluid percussion injury; PBBI, penetrating ballistic-like brain injury.

Fluid percussion injury model

A group effect on GFAP levels was detected at 4 (p < 0.0001) and 24 h (p < 0.0001), with all injured groups showing significant increases versus sham (Fig. 4A). GLY-treated rats also had significantly higher GFAP levels than FPI vehicle (p = 0.04) at 4 h after injury. However, the trend toward increased GFAP at 24 h after injury in the GLY-treated group versus vehicle did not reach significance, and thus no points were generated for the OBTT scoring matrix (Fig. 4A), given that the 24-h biomarker levels are the outcome that is scored in OBTT. Similarly, no significant difference on delta 24–4 h GFAP levels was found (Fig. 5A).

At 4 h post-injury, relative to sham, all FPI groups exhibited significant increases in levels of UCH-L1 (p < 0.001), but there was no difference between vehicle- and GLY-treated groups. In contrast, no significant between-group effects for any FPI group versus sham were observed at 24 h (Fig. 4A). Delta 24–4 h UCH-L1 levels showed no treatment effect (Fig. 5A). Thus, no points were generated for the OBTT matrix for UCH-L1 in FPI.

Controlled cortical impact model

Compared to shams, serum GFAP concentrations were substantially increased in both CCI groups at 4 (p < 0.0001) and 24 h (p = 0.0002) post-injury (Fig. 4B). Although, at 24 h, GFAP levels were ∼3-fold lower in rats treated with GLY than their respective vehicle controls, this difference did not reach significance, and also delta 24–4 h GFAP levels did not differ between vehicle- and GLY-treated groups (Figs. 4B and 5B). Thus, no points were generated for GFAP in the CCI model in the OBTT scoring matrix.

There were no significant group effects on post-injury levels of UCH-L1 at 4 or 24 h (Fig. 4B) or on delta 24–4 h UCH-L1 levels (Fig. 5B). Thus, no points were generated for UCH-L1 in the CCI model in the OBTT scoring matrix.

Penetrating ballistic-like brain injury model

All injured groups showed significantly increased levels of GFAP versus sham at both time points, with lower levels in GLY- versus vehicle-treated rats (>1.5-fold lower) at 24 h. However, once again, these differences did not reach statistical significance (Fig. 4C). Also, no significant difference was detected between the injured groups on delta 24–4 h GFAP levels (Fig. 5C). Thus, no points were generated for GFAP in the PBBI model in the OBTT scoring matrix.

The overall analysis revealed a significant main effect on UCH-L1 levels at 4 (p = 0.01) and 24 h (p = 0.002) post-injury. Serum levels of UCH-L1 were significantly higher in both PBBI groups versus sham, but there was no difference between the injured vehicle and treatment groups (Fig. 4C). Delta 24–4 h UCH-L1 levels also did not differ between vehicle- and GLY-treated rats (Fig. 5C). Thus, no points were generated for UCH-L1 in the PBBI model in the OBTT scoring matrix.

Operation Brain Trauma Therapy outcome scoring matrix

The overall scoring matrix is shown in Table 3 for the effect of GLY across models. Overall, GLY was most beneficial in CCI receiving a +4 points for motor benefit and +2 points for neuropathological benefit. However, the total score for CCI was +3.5 because of the score of −2.5 from worse performance on the MWM hidden platform latency in the GLY-treated group. GLY was also beneficial in FPI receiving a total of +2 points awarded for motor benefit, as measured by the cylinder task. The greatest benefit of GLY across models was on motor function, with a total of +5.5 points (+2 in FPI and +3.5 in CCI). There were no benefits observed with GLY in PBBI.

Morbidity and mortality

No adverse effects or apparent acute physiological problems from GLY were observed in the FPI model. No notable mortality from GLY was appreciated in FPI, or CCI, although as indicated previously, after CCI, 3 rats in the vehicle and 1 in the GLY groups could not swim in the MWM task. In PBBI, 2 vehicle rats died versus 1 GLY-treated rat. Taken together, these observations indicate no detrimental impact of GLY treatment on mortality or morbidity.

Ancillary studies

Glibenclamide pharmacokinetics

Mean serial plasma GLY levels were 5.42 ± 0.63 pg/mL at 1 h, 3.16 ± 0.40 pg/mL at 2 h, 2.44 ± 0.42 pg/mL at 4 h, 1.04 ± 0.16 pg/mL at 20 h, and 1.23 ± 0.11 pg/mL at 24 h after bolus administration. Post-load and steady-state levels were thus 5.42 and 1.23 pg/mL, respectively. In 2 rats in which plasma and CSF were simultaneously assessed at 24 h after CCI, (using the identical injury and treatment parameters as in the outcome study), CSF levels ranged between 22% and 50% of plasma level.

Blood glucose levels

Blood glucose levels (mg/dL) in sham rats treated with vehicle versus GLY were assessed at 1, 2, 4, 20, and 24 h after treatment (bolus plus infusion). Values were 136.8 ± 3.5 versus 159.8 ± 14.75 at 1 h, 146.4 ± 15.5 versus 134.6 ± 9.2 at 2 h, 128.8 ± 3.2 versus 143.6 ± 3.8 at 4 h, 159.6 ± 32.6 versus 162.0 ± 24.3 at 20 h, and 133.2 ± 6.2 versus 124.4 ± 4.9 mg/dL at 24 h in vehicle- versus GLY-treated groups, respectively. The groups did not significantly differ at any time point. There were also no episodes of hypoglycemia. Specifically, no glucose value was <113 mg/dL.

Cerebral edema assessment

Given the purported beneficial effects of GLY on cerebral edema in the CNS by Sur1-Trpm4 channel inhibition, and its performance in the CCI model as discussed above, in separate studies, brain edema was assessed at 24 h post-CCI using the wet-dry method determining hemispheric %BW measurement in three groups: naïve, CCI-vehicle, and CCI-GLY (n = 5 per group). In naïve rats, %BW was 78.88 ± 0.04% ipsilaterally and 78.90 ± 0.03% contralaterally, reflecting normal values. At 24 h post-CCI, ipsilateral %BW in the vehicle group was increased at 80.36 ± 0.12%, with no change in the contralateral %BW (78.7 5 ± 0.03%). Ipsilateral %BW remained increased at 24 h in GLY-treated rats (80.63 ± 0.36%), with no change in contralateral %BW, which remained normal at 78.73 ± 0.04%. Ipsilateral hemispheric %BW in both groups of injured rats (CCI-vehicle and CCI-GLY) was increased versus naive (p < 0.05, ANOVA, SNK), and, surprisingly, this ∼2% increase in hemispheric brain water was not attenuated by GLY (post hoc SNK).

Discussion

In this OBTT consortium study, treatment with GLY showed benefit in FPI and CCI, but not PBBI. Of the 12 drugs tested by OBTT, GLY is the second-highest–scoring drug after levetiracetam with notable features.⁶¹ The greatest impact of treatment was identified in CCI where GLY improved both motor function and 21-day lesion volume. It is the only OBTT drug to date (of a total of 12 tested) to significantly reduce lesion (contusion) volume in CCI, consistent with earlier pre-clinical and clinical reports.^17,21,34,36 However, the benefit of GLY in FPI was limited to motor function. GLY did not improve cognitive function in any of the models, but instead received partial negative points (−2.5 of the 5 available) for an observed MWM deficit in CCI with treatment—when a statistically significant deficit was not observed in the CCI-vehicle group. There was also no effect of GLY on either blood biomarker (GFAP or UCH-L1) in any model, although a non-significant 3-fold reduction in GFAP levels for treated versus control was noted at 24 h after injury in CCI.

Our dosing protocol achieved levels of GLY fairly consistent with earlier reports, with substantial brain exposure after injury. We also did not observe any episodes of hypoglycemia, consistent with the increased sensitivity of CNS SUR1-TRPM4 to lower doses of GLY versus the K_ATP channels in pancreatic beta cells. We observed no mortality or morbidity concerns with the drug, supporting safety, although a modest (∼9 mm Hg) decrease in MAP was observed in the FPI model at 5 min post-bolus (15 min post-injury) in GLY-treated rats. Each of these points merits discussion.

Glibenclamide and contusion expansion

The reduction in contusion volume in CCI by GLY in this blinded, rigorous OBTT study mirrors earlier reports and supports ongoing clinical studies in contusion.^17,21,33,36 Contusion expansion occurs in 30–50% of patients.^62–67 Hemorrhage progression of contusion predicts unfavorable clinical outcome.^68–70 Thus, limiting contusion growth could potentially improve patient outcomes.

Curbing of contusion expansion by GLY may result from inhibition of SUR1-TRPM4 in cerebrovascular endothelial cells.^16,47,64,71 Cerebrovascular endothelial SUR1-TRPM4 expression is upregulated after injury and contributes to edema in these cells. As the endothelial cells continue to swell, and tight junctions between them are degraded, the capillaries become “fenestrated” and proteinaceous fluid extravasates, furthering vasogenic edema. In its extreme form, there is a complete loss of BBB integrity, resulting in progressive hemorrhage and contusion expansion.^{16,47,64,71,72} This pathway may also overlap with matrix metalloproteinase-9 (MMP-9), a known contributor to BBB breakdown and hemorrhage progression.^16,24,73 After contusion in rats, SUR1 upregulation in capillary endothelium is evident by 3 h and is prominent in the cortex, hippocampus, and thalamus by 24 h.²¹ By 72 h, SUR1 is predominantly expressed in small round cells (thought to be macrophages).¹⁷ In human contusion, elevated SUR1 expression was observed by 12 h post-TBI and was sustained.⁴⁶ Colocalization of SUR1-TRPM4 in rat microvessels is evident in core contusion.¹⁷ Colocalization (immunolabeling) and coassembly (fluorescence resonance energy transfer) of microvascular SUR1-TRPM4 has also been shown in human contusion.¹⁷

Previous reports of GLY treatment in various rodent versions of CCI (4.5 mm depth × 1 m/s velocity; 1.5 mm depth × 1.5 m/s velocity; 1.5 mm depth × 7.5 m/s) or weight drop (10-g weight × 5 cm drop length) consistently show reduced BBB permeability and decreased contusion volumes at multiple time points (from 45 min to 72 h).^17,21,33,34 We quantified lesion volume and benefit of GLY at 21 days (vs. more acute time points in the literature), expanding on earlier reports and suggesting an enduring benefit. The consistency of GLY benefit on contusion size in CCI, regardless of injury severity and in multiple species, may increase the probability of successful clinical translation. The lack of effect on lesion volume in the other models that we tested, however, identifies contusion as a mechanistic target for GLY, suggesting that the direct comparison across different injury phenotypes is a valuable facet of the OBTT strategy, with its direct cross-model comparisons. Although parasagittal FPI, at the injury level used in OBTT, has a contusional component, it reflects modest tissue disruption at the gray-white interface. In contrast, the CCI injury level used in OBTT results in substantial hemorrhagic parenchymal and vascular disruption in the cortex, which may be essential for robust upregulation of Sur1-TRPM4 and progressive expansion.

The preliminary clinical studies of GLY in TBI are small, although they align with pre-clinical findings, and indicate that GLY reduces contusion volume (and/or rate of expansion). A randomized trial of 66 patients with moderate-severe TBI and contusion volumes <30 mL found that treatment with oral GLY (10 mg/day × 10 days) decreased contusion expansion rate.³⁶ The recent pilot phase II study of intravenous GLY identified a trend toward benefit by treatment.³⁴ The extremely small sample size (n = 7 per group) was not powered for efficacy; however, contusion growth was 10-fold higher in placebo versus GLY (1036 ± 1963.28% vs. 136 ± 195.62%; p = 0.15). In that study, three magnetic resonance imaging (MRI) measures of edema (free water, mean diffusivity, and tissue mean diffusivity) in lesions versus uninjured white matter were different only in the placebo arm (p < 0.02). Nonetheless, caution is in order with interpreting these preliminary findings. Leveraging the mounting evidence for GLY benefit on contusion (including our results), ASTRAL (NCT03954041) is an ongoing international multi-center, double-blind, randomized phase II study focused on brain contusion evaluating whether intravenous GLY reduces radiographical 96-h expansion versus placebo. The estimated trial enrollment is 160 patients.

Post-traumatic cerebral edema has both cytotoxic and vasogenic components that may be distinguishable on MRI.^74–77 Given the upregulation of SUR1-TRPM4 in multiple CNS cell types, it has been implicated in both cytotoxic and vasogenic edema.¹⁶ GLY may thus differentially affect edema subtypes based on time, dose, and regional expression differences. Exploring more nuanced regional and cellular effects of GLY on edema subtypes and BBB permeability may provide further insights into the pathobiological basis underlying the different treatment responses across models, to pave the way for trials tailored to specific patient/injury characteristics.

Glibenclamide and cognitive outcome

GLY failed to produce any benefit on any cognitive outcome parameter across our models and produced −2.5 points in CCI. The negative points generated by GLY for cognitive outcome (MWM latency testing) in CCI were unexpected and discordant with the one other study in CCI evaluating the effects of GLY on cognitive outcome.³² In that study, GLY-treated rats showed no difference versus vehicle or sham in MWM latency, but had improved probe-trial performance. Whereas the dose of GLY was the same as in our study, the injury was less severe than in OBTT. Although model differences might explain the discrepancy between these studies, other considerations could have influenced the findings. We did not observe significant MWM deficits in vehicle-treated injured rats in either the FPI or CCI model. This limited our ability to gauge the therapeutic effect. That may have biased against GLY, given that the number of potential points that it could have achieved was restricted versus many of the other drugs tested by OBTT. We also noted that in CCI, three of the vehicle-treated versus one of the GLY-treated rats could not swim and thus were not scored in the MWM—as defined in the pre hoc criteria for performance testing on this task; that may have further biased against GLY if persistent motor deficits reflected greater overall deficits in those rats.

Similarly, no therapeutic target was observed for either FPI or CCI in the probe trial. The importance of this limitation is magnified by the fact that cognitive outcome is the parameter with the largest number of potential points for therapies tested in OBTT. This concern has occurred sporadically in FPI and CCI in earlier OBTT reports. There is a difficult balance in selection of injury level between achieving a robust deficit on a given task versus producing an insult of a magnitude that is refractory to any therapy. Nevertheless, a true detrimental effect on cognitive outcome by GLY cannot be ruled out, and no trends toward cognitive benefit were noted on any of the tasks across the OBTT models. Lack of cognitive benefit was disappointing, although the number of cognitive tasks assessed by OBTT is limited, given the screening nature of the approach in OBTT. Further studies are warranted.

Biomarker response

In our earlier studies of the first five therapies tested by OBTT, GFAP showed promise as a marker to assess response to a therapy and its relationship to a conventional outcome parameter. In our OBTT report on levetiracetam, reduced blood GFAP level at 24 h in CCI predicted response to the drug (i.e., a reduction in hemispheric tissue loss at 21 days after TBI).⁶¹ In the current study, GFAP blood levels at 24 h showed declines in values toward sham levels in the GLY-treated group in CCI, though remaining elevated. This trend would need to be further explored in a larger sample to appropriately test the hypothesis that 24-h blood GFAP levels predict treatment efficacy. Nonetheless, the strong correlation between acute circulating GFAP concentrations and contusion volume characterized in a previous OBTT investigation⁷⁸ is once again supported in this study (Supplementary Fig. S1), providing added evidence that 24-h blood levels of GFAP could serve as an early and convenient screening tool to predict histological end-points, which are very labor-intensive to assess.

Because of its rapid decay in blood across all rat models, UCH-L1 has not proven to be a useful biomarker of treatment response in OBTT. In this study, UCH-L1 increases were again not observed at 24 h versus sham in FPI or CCI. Conversely, UCH-L1 was increased at both 4 and 24 h after TBI versus sham in PBBI—possibly reflecting the severity and ongoing tissue injury in that model. Still, no drug treatment effects were observed. Last, in future studies, it might be useful to introduce additional TBI biomarkers as end-points. The axonal injury-tracking neurofilament proteins, such as neurofilament-L and phosphorylated neurofilament-H, are gaining momentum as clinical TBI diagnostic and prognostic biomarkers^79–81 and as monitoring biomarkers in pre-clinical TBI as well as spinal cord injury.^9,82,83

We found correlation disparities between biomarkers and the other outcomes across models (Supplementary Fig. S1). This suggests phenotype-dependent pathobiological characteristics and mechanisms that contribute to biomarker concentrations and dynamics.^84–86 Understanding how these factors act and interact to influence biomarker release and kinetics may reveal new insights into TBI heterogeneity and could inform future diagnostic and therapeutic advances.

Glibenclamide pharmacokinetics

Concerns have been raised surrounding the use of literature-based dosing to guide protocol design, rather than assessing drug levels, brain pharmacodynamics, and/or target engagement. In this study, the first in the second series of studies by OBTT (i.e., drug no. 6), we attempted to address this concern within the limits of a multi-center drug-screening consortium. First, we used a treatment protocol that had been published, and we carried out studies to measure both plasma and CSF levels of GLY in separate rats. Previous studies in rats suggest that an i.p. load of 10 μg/kg, followed by 0.2 μg/h of GLY by s.c. infusion, yields a plasma level of ∼5 ng/mL.⁷⁸ We observed a peak plasma level of 5.42 ng/mL at 1 h and 1.23 ng/mL at 24 h, along with confirmation of a substantial brain exposure, with CSF levels between 22% and 50% of plasma, in a small exploratory study.

This builds on the earlier work, but questions remain given that our pharmacokinetic studies were restricted to CCI. Also, without confirmation of target engagement, it is unclear what the optimal dose and therapeutic window are in our models. Given differences in BBB permeability and brain perfusion after injury across models, different doses and/or duration of therapy might be needed in each model—particularly given that GLY is not expected to cross the intact BBB. Indeed, given the findings in CCI, and the fact that there is an ongoing clinical trial, we believe that a dose-response study is warranted in each model.

Glibenclamide and cerebral edema

The lack of treatment effect on 24-h cerebral edema as measured by %BW was surprising and contrasted with earlier reports.^33,39,44 There are several notable factors about this finding in the context of the literature. Aside from model and species differences versus OBTT, Xu and colleagues treated mice post-CCI with high-dose GLY (10 μg/d × 3 days) and identified reduced %BW at 72 h (79.7% vs. 81.3% in vehicle; p < 0.05).³³ However, Zweckberger and colleagues used the same dose of GLY as OBTT, initiated s.c. at 15 min after CCI, and the reduction in %BW versus placebo was smaller (80.47 ± 0.37% vs. 80.83 ± 0.44%) and required 14 mice per group to achieve significance. The 24-h measurement in OBTT may have been premature and the lack of effect dose and/or duration related. It is also possible that the impact of GLY on lesion (contusion) volume and motor function is by means of mechanisms independent of cerebral edema formation (e.g., MMP-9), or that drug effect on endothelial cells is not affected by BBB permeability/CNS availability and is thus more pronounced in these cells versus astrocytes/neurons that require CNS penetration. It is also plausible that %BW, though often viewed as a gold standard, fails to measure local differences in edema or differentiate their subtypes. In any case, we were surprised to observe no visible trend for GLY on edema in CCI given the effect on contusion volume.

Limitations

Although the strengths of the approach used by OBTT have been previously discussed, there are limitations to multi-center screening and to this assessment of GLY. We used a single injury level in each model, generally representing moderate-to-severe injury. This allowed us to compare findings with our earlier work; the same models and injury levels were used in all studies.^{49,61,87–90} However, various drugs could have different effects at different injury levels. We also did not explore sex differences and/or assess other outcomes, such as hippocampal neuronal death. These should be pursued for GLY, given that it has shown some promise in OBTT. However, these studies are beyond the scope of a screening consortium gauging potential drug efficacy and disparities across models using fixed dosing regimens.

We also only examined the effect of treatment on brain edema in CCI and by a single method. That decision was made post hoc and is not part of the scoring matrix. Those studies were ancillary and were carried out after the primary outcome codes had been broken across sites for GLY, which revealed a reduction in lesion volume in CCI. Our studies of brain edema were done with the same level of rigor as the primary outcome studies, but were restricted to CCI in an attempt to better understand the potential mechanism of action of GLY and the scope of its effect in the model where it showed greatest efficacy.

A secondary mandate of OBTT is to carry out additional studies of promising drugs to guide clinical trials. Thus, our focus is on CCI for the ancillary studies. Finally, our studies are not sufficiently powered for many of the outcomes tested and that, in some cases, the therapeutic target for a given outcome was not achieved at the injury level used. In this case, our design was biased somewhat against GLY, as discussed earlier. Finally, in all three OBTT models, a craniotomy was performed to induce TBI. Given that GLY might reduce brain swelling/intracranial pressure, benefit could be underestimated by the craniotomy, distinct from the condition without surgical treatment.

Conclusions

GLY is the second-highest–scoring agent tested to date by OBTT, after levetiracetam. It had positive effects in two models and across multiple outcomes. It improved motor outcome in both FPI and CCI, and is the only drug evaluated by OBTT to reduce 21-day lesion (contusion) volume after CCI. The lack of benefit on cerebral edema post-CCI and cognitive outcome in any of the three models was unexpected and contrary to reports in the literature. This may reflect a true lack of benefit or be related to study design, including dose, duration of therapy, sensitivity of the specific outcome measures, or bias against treatment related to limited insult severity in two of the models. GLY was safe and did not cause hypoglycemia. The clear reduction in contusion volume is consistent with multiple earlier studies in CCI, despite severity, dose, and species differences, and, along with safety, supports the potential for clinical translation and/or additional testing in small and large animal models. GLY is being evaluated in the ongoing phase II ASTRAL trial in brain contusion. Optimizing treatment regimens (dose, duration, and timing), surrogate markers for edema subtypes on MRI, pathway-specific biomarkers, and genetic risk stratification may facilitate precision medicine and patient selection for future clinical trials.

Authors' Contributions

All of the authors were involved in the conception or design of the work, critical revision of the article, and final approval. Drs. Jha, Mondello, Bramlett, Dixon, Shear, Wang, Yang, Poloyac, Yan, Carlson, and Kochanek were involved in data collection, analysis, and interpretation and in drafting the article.

Funding Information

We are grateful to the U.S. Department of Defense grants WH81XWH-10-1-0623 and WH81XWH-14-2-0018 for generous support. Dr. Ruchira M. Jha was supported by NIH NINDS K23NS101036 grant.

Author Disclosure Statement

Dr. Hayes owns stock in and is an officer of Banyan Biomarkers Inc. Dr. Hayes is an employee of and receives salary and stock options from Banyan Biomarkers Inc. Dr. Wang is a former employee of Banyan Biomarkers Inc. and owns stock. Drs. Hayes and Wang also receive royalties from licensing fees, and, as such, these individuals may benefit financially from the outcomes of this research or work reported in this publication. Dr. Jha is a paid consultant for Biogen and is on its advisory board.

Supplementary Material

Supplementary Figure S1