Cerebrospinal Fluid Sulfonylurea Receptor-1 is Associated with Intracranial Pressure and Outcome after Pediatric TBI: An Exploratory Analysis of the Cool Kids Trial

Abstract

Sulfonylurea receptor-1 (SUR1) is recognized increasingly as a key contributor to cerebral edema, hemorrhage progression, and possibly neuronal death in multiple forms of acute brain injury. SUR1 inhibition may be protective and is actively undergoing evaluation in Phase-2/3 trials of traumatic brain injury (TBI) and stroke. In adult TBI, SUR1 expression is associated with intracranial hypertension and contusion expansion; its role in pediatric TBI remains unexplored. We tested 61 cerebrospinal fluid (CSF) samples from 16 pediatric patients with severe TBI enrolled in the multicenter Phase-3 randomized controlled “Cool Kids” trial and seven non-brain injured pediatric controls for SUR1 expression by enzyme-linked immunosorbent assay. Linear mixed models evaluated associations between mean SUR1 and intracranial pressure (ICP) over the first seven days and pediatric Glasgow Outcome Scale-Extended (GOS-E Peds) over the initial year after injury. SUR1 was undetectable in control CSF and increased versus control in nine of 16 patients with TBI. Mean SUR1 was not associated with age, sex, or therapeutic hypothermia. Each 1-point increase in initial Glasgow Coma Score was associated with a 1.68 ng/mL decrease in CSF SUR1. The CSF SUR1 was associated with increased ICP over seven days (b = 0.73, p = 0.004) and worse (higher) GOS-E Peds score (b = 0.24, p = 0.004). In this exploratory pediatric study, CSF SUR1 was undetectable in controls and variably elevated in severe TBI. Mean CSF SUR1 concentration was associated with ICP and outcome. These findings are distinct from our previous report in adults with severe TBI, where SUR1 was detected universally. SUR1 may be a viable therapeutic target in a subset of pediatric TBI, and further study is warranted.

Introduction

Injury related deaths are the most common cause of mortality in children and adolescents in the United States.¹ Although severe traumatic brain injury (TBI) is a small fraction of pediatric TBI, it results in a disproportionate degree of death and disability.^2,3 Therapeutics in pediatric severe TBI have been understudied, but this landscape is changing: the recently completed Approaches and Decisions in Acute Pediatric TBI Trial (ADAPT) enrolled 1000 children and may provide insight into current therapy.⁴ In studies of large adult TBI cohorts, initial injury characteristics only explain ∼35% of final outcome variability.⁵ The remainder is thought to be influenced by secondary injury mechanisms such as cerebral edema, hemorrhage progression, and intracranial hypertension.

The 2019 guidelines for severe pediatric TBI and accompanying algorithm appropriately address intracranial pressure (ICP) monitoring, thresholds, and management strategies.^6,7 Understanding the molecular underpinnings of these processes is critical to therapeutically targeting early molecular drivers of injury and improving functional outcome in both adults and children.⁸

Numerous pre-clinical and clinical studies suggest that the sulfonylurea receptor-1 (SUR1)-transient receptor potential cation channel subfamily M member 4 (TRPM4) non-specific monovalent cation channel is a key mediator of secondary injury after TBI.⁹ Specifically, previous research has demonstrated that SUR1-TRPM4 plays an important role in post-traumatic cerebral edema, contusion expansion, and intracranial hypertension.^10–14 Further, SUR1, the regulatory component of this heteromeric channel, is inhibited by glyburide, a familiar sulfonylurea antidiabetic medication.¹⁵

Current evidence suggests that glyburide may limit cerebral edema development and is potentially neuroprotective after TBI; an intravenous formulation of glyburide is being evaluated in a Phase-II trial in contusional adult TBI (NCT03954041), based on the results of a previous Phase-II trial in TBI demonstrating a possible benefit specifically in contusional TBI,¹⁶ and a Phase-III trial in adult ischemic stroke (NCT02864953).^16,17

In adults, we reported that SUR1 is undetectable in the cerebrospinal fluid (CSF) of controls, but markedly upregulated in all patients after severe TBI.¹⁸ SUR1 was higher in adults with radiographic evidence of cerebral edema. The role of this channel in pediatric TBI remains unexplored, however.

The purpose of this exploratory study was to assess whether SUR1 was detectable after severe pediatric TBI and if it was associated with ICP or functional outcome. Based on our previous findings in adults, we hypothesized that (1) SUR1 is detectable in the CSF of pediatric patients with severe TBI but not from non-head injured pediatric controls, (2) Increased SUR1 is associated with increased ICP, and (3) increased SUR1 is associated with unfavorable outcomes as measured by the pediatric Glasgow Outcome Scale-Extended (GOS-E Peds) score.

Methods

Participants

Patients with TBI had been enrolled in the Cool Kids Trial of hypothermia in severe pediatric TBI as described previously.¹⁹ All patients' healthcare proxies provided consent for the trial and CSF collection. Briefly, pediatric (age <18) patients with severe TBI (Glasgow Coma Scale [GCS] score ≤8) were randomized to 48–72 h of hypothermia (32–33°C) or normothermia (36.5–37.5°C) within 6 h of injury. For this study, all CSF samples from each study patient enrolled at our site with at least one CSF sample were included, for a total of 54 samples. Deidentified control CSF from lumbar punctures in seven non-head injured patients, containing 0 red blood cells and 0 white blood cells, was obtained from the Children's Hospital of Pittsburgh clinical laboratory. The University of Pittsburgh Institutional Review Board approved this study.

CSF collection and SUR1 quantification

The CSF was collected from external ventricular drains between 12 and 137 hours post-injury (median 40 h) as clinically indicated per TBI standard of care and immediately placed on ice. Samples were centrifuged for 8–10 min at 5000g as soon as possible and stored at -70°C until analysis without interval freeze-thaw. The SUR1 from undiluted CSF was quantified using enzyme-linked immunosorbent assay (ELISA, MBS2000195, MyBioSource, San Diego, CA). Assays were performed by technicians blind to demographics and outcomes. Assay lower limit of detection was 0.117 ng/mL, and SUR1 level of any sample below the detection limit was recorded as 0 ng/mL. Because of limited sample volumes, samples were run singly on a single plate.

Patient demographics and clinical outcomes

Patient age, sex, admission GCS score, mechanism of injury, hourly ICP measurements over the first 178 h after injury, and GOS-E Peds at discharge, 3, 6, and 12 months were obtained from the Cool Kids Trial database by research coordinators blind to SUR1 concentrations. The coprimary outcomes were ICP over seven days and GOS-E Peds over 12 months. The GOS-E Peds, as described in the Cool Kids trial,¹⁹ is an 8 point ordinal scale where, opposite the adult scale, lower scores represent better outcomes (1 is no disability, 8 is dead).

Statistical analysis

The SUR1 values are reported as both the mean CSF SUR1 level across serial samples tested for each patient and the peak SUR1 level for that patient. We tested for an association between SUR1 level and time from injury by both linear regression and mixed effects models to look for a temporal trend in SUR1 expression. Descriptive summaries of continuous and ordinal variables are presented as median (interquartile range [IQR]); categorical variables are provided as fractions and percentages. Based on normality, the association between age or admission GCS score and mean SUR1 was tested using simple linear regression, while association between sex or therapeutic hypothermia and SUR1 was tested using Wilcoxon rank-sum tests.

The primary analysis utilized linear mixed effects random coefficient (RC) models to test the association between mean SUR1 and repeated measures of ICP and GOS-E Peds. Likelihood ratios tests demonstrated that models including quadratic outcome-time association were superior to those including only linear outcome-time association, and that models with an unstructured covariance matrix outperformed models without slope-intercept covariance. Therefore, final models included both a time-squared term and an unstructured covariance matrix. Because of limited sample size, the primary analysis included only variables for time and mean SUR1 to minimize model overfitting.

Secondary exploratory analyses included RC models adjusted for age, sex, and GCS, RC models adjusted for interventional arm, RC models using peak, rather than mean, SUR1, RC models assessing a SUR1 by time interaction, RC models where SUR1 was dichotomized as detectable in any sample or never detectable, and simple linear regression models testing associations between SUR1 and mean ICP or three-month GOS-E Peds.

Alpha was set at 0.05 given the study's exploratory nature and known association between ICP and outcome. All p values are two-sided. Analyses were performed using Stata SE 16.1 (StataCorp, College Station, TX).

Results

Demographic and clinical data for patients with TBI are presented in Table 1. Nine of 16 (56.2%) patients were male, median GCS score was 7 (5.5–7), and median age on admission was 10.8 (2.3–13.8) years. Eight of 16 (50%) received therapeutic hypothermia. Only one patient (6.3%) had died by 12 months.

Table 1.

Demographic and Clinical Characteristics of Included Pediatric Patients with Traumatic Brain Injury

Patient	Age (years)	Sex	Admission GCS	MOI	Interventional arm	Peak CSF SUR1 (ng/mL)	Mean CSF SUR1 (ng/mL)	Mean ICP (mmHg)	12 month GOS-E – Peds
1	13.5	F	5	Fall	Hypothermia	0	0	9.7	2
2	14.0	M	7	RVA	Normothermia	0.308	0.123	15.2	3
3	5.7	M	8	MVP	Hypothermia	0	0	7.1	1
4	14.7	M	7	Other	Hypothermia	0.270	0.068	7.1	1
5	13.9	M	7	RVA	Normothermia	0	0	9.1	2
6	12.0	F	7	Other	Normothermia	0.270	0.068	14.7	1
7	14.4	M	7	RVA	Normothermia	0	0	12.1	2
8	3.3	F	7	Fall	Normothermia	0.379	0.095	12.8	6
9	11.4	M	5	MVA	Hypothermia	0.284	0.188	5.3	2
10	14.7	F	4	Other	Hypothermia	17.784	13.619	19.1	8
11	0.8	M	7	RVA	Hypothermia	0.360	0.180	7.5	2
12	1.2	M	6	RVA	Hypothermia	0	0	8.9	3
13	1.4	F	5	MVA	Normothermia	0.657	0.219	8.8	3
14	0.8	F	7	Other	Normothermia	0	0	8.0	1
15	10.8	F	8	Other	Hypothermia	0	0	7.2	1
16	2.1	M	6	Other	Normothermia	0.544	0.109	14.0	6

GCS, Glasgow Coma Scale; MOI, mechanism of injury; CSF, cerebrospinal fluid; SUR1, sulfonylurea receptor 1; ICP, intracranial pressure; GOS-E Pediatric, Glasgow Outcome Scale–Extended, pediatric revision; RVA, recreational vehicle accident; MVP, motor vehicle vs. pedestrian; MVA, passenger in motor vehicle; Other includes kicked by horse, motor vehicle vs. horse buggy, and bicycle accident

CSF SUR1 quantification

We tested 54 CSF samples from pediatric patients with TBI, of whom 14 had measurable SUR1. Nine of 16 (56.3%) patients with TBI had detectable SUR1 versus 0/7 controls (p = 0.019). The cohort median of each patient's mean CSF SUR1 level across samples was 0.068 (0–0.152) ng/mL overall, 0.123 (0.095–0.188) ng/mL among those with detectable SUR1. There was no association between CSF SUR1 level and time from injury to sample collection using either linear regression or a RC model. Mean SUR1 was not associated with age (b = 0.16, p = 0.30) or sex (male: 0.068 [0–0.123] ng/mL vs. female: 0.068 [0–0.219 ng/mL], p = 0.85). SUR1 levels did not differ by normothermia versus hypothermia (0.081 [0–0.116] ng/mL vs. 0.034 [0–0.184] ng/mL, respectively, p = 0.97). The GCS score was inversely associated with SUR1 (b = -1.68, p = 0.021), or a 1.68 ng/mL decrease in CSF SUR1 for each 1 point increase in GCS.

Association of mean SUR1 with ICP and GOS-E Peds

Increased mean SUR1 was associated with higher ICP over seven days (b = 0.73, p = 0.004) and worse functional outcome on the GOS-E Peds scale (b = 0.24, p = 0.004) over 12 months. In models adjusting for age, sex, and admission GCS, higher SUR1 remained associated with increased ICP (b = 0.71, p = 0.027) and worse GOS-E Peds (b = 0.19, p = 0.05). SUR1 remained associated with GOS-E Peds after adjustment for therapeutic hypothermia. CSF SUR1 also remained associated with increased ICP after adjusting for hypothermia (b = 0.84, p < 0.001), which was itself associated with decreased ICP (b = -5.53, p < 0.001). Increased SUR1 was also associated with increased mean ICP (b = 0.69, p = 0.012) and unfavorable GOS-E Peds scores at three months (b = 0.36, p = 0.024). Similar to mean SUR1, higher peak SUR1 was associated with higher ICP (b = 0.57, p = 0.003) and higher GOS-E Peds score (b = 0.19, p = 0.003).

There was no evidence of a SUR1 by time interaction in association with ICP (p_{LR test} = 0.34); i.e., increased SUR1 did not associate with the rate of ICP change over time. There was evidence of this interaction, however, in association with GOS-E Peds (p_{LR test} = 0.049)—patients with undetectable SUR1 had significant improvement in outcome over time (b = -0.81), but the rate of improvement decreased as SUR1 increased (b_interaction = 0.02).

In patients with SUR1 detected in any CSF sample, there was a trend toward increased ICP over seven days when compared with those with no detectable SUR1 in any sample (b = 3.23, p = 0.084, Fig. 1A). Given the underlying biology of SUR1 upregulation, we explored this relationship beyond 24 h from injury: when limited to this time frame, measurable SUR1 was associated with higher ICP values (b = 4.04, p = 0.034, Fig. 1A). Detectable SUR1 was also associated with worse outcomes over 12 months (b = 1.17, p = 0.045, Fig. 1B).

FIG. 1.

Association between detectable sulfonylurea receptor 1 (SUR1) and outcome. (A) Detectable cerebrospinal fluid (CSF) SUR1 was associated with a trend toward increased intracranial pressure (ICP) over the first seven days after injury (p = 0.084). This becomes significantly associated with increased ICP after the early post-injury period (24 h, shown by the vertical dashed line in the figure) during purported SUR1 up-regulation (p = 0.034). Error bars represent 95% confidence intervals. (B) Detectable CSF SUR1 was associated with worse functional outcome on the Glasgow Outcome Scale-Extended Pediatrics (GOS-E Peds) over 12 months (p = 0.045). Shaded areas represent 95% confidence intervals.

Discussion

In this exploratory study of pediatric patients with severe TBI from the multi-center Cool Kids hypothermia randomized controlled trial (RCT), CSF SUR1 was variably detected after TBI, but undetectable in all controls. Higher levels of SUR1 were associated with increased ICP over the first seven days after injury and with worse functional outcomes over 12 months.

These results are consistent with our previous report in adult TBI, where CSF SUR1 was undetectable in controls but measurable after injury.¹⁸ In our pediatric sample, however, SUR1 was only detectable in 56.3% of patients with TBI, and even when CSF SUR1 was detectable, the levels were much lower in pediatric patients than in adults (0.123 vs. 3.54 ng/mL).¹⁸ This may reflect important differences between pediatric versus adult pathophysiology and underlying mechanistic involvement of SUR1-TRPM4 as it contributes to secondary injury such as cerebral edema, intracranial hypertension, and hemorrhage progression, or developmental differences in SUR1 expression. It may also be because of differences in CSF processing, however. Of note, samples from our report in adults were not collected within the context of a RCT.

Heterogeneous SUR1 expression may also relate to genetic variation—previous reports demonstrate strong associations between ABCC8 polymorphisms (the gene encoding SUR1), ICP, and outcome after adult TBI.^11,20–22 Differences in epigenetics and expression regulation may also contribute, but have not currently been reported. Despite more limited SUR1 expression in these pediatric TBI samples, CSF SUR1 levels were associated with both mean ICP and three-month GOS-E Peds, whereas this was not the case in adult patients.¹⁸ These findings are consistent with underlying biology—studies in some rodent TBI models demonstrate reduced brain water content (a pre-clinical marker for cerebral edema) in animals treated with the SUR1 inhibitor glyburide.^13,23

Interestingly, SUR1 expression did not differ between patients randomized to hypothermia versus normothermia. Although limited by sample size, this suggests the possibility that hypothermia may influence intracranial hypertension through mechanisms independent of SUR1. Further, SUR1 remained significantly associated with increased ICP after adjusting for hypothermia, which was itself associated with decreased ICP, suggesting complementary mechanisms. These results contrast with those of a pre-clinical cardiac arrest study finding no additive effects of SUR1 inhibition and hypothermia.¹⁴

Although this study is strengthened by the fact that randomization, protocolized clinical care, and CSF sampling were performed in the context of an RCT, our study has several limitations—the most prominent being our sample size of 16 patients and seven controls. Nonetheless, we had several CSF samples at multiple time points (n = 61), and our findings were robust to covariate adjustment, remained significant after dichotomization of SUR1 values, and are consistent with previous reports and underlying pathophysiology. Nevertheless, we recognize that it is exploratory.

Unfortunately, only limited data were available on the head CT scans of these patients in the Cool Kids database, given that the trial began recruitment in 2007. Thus, we could not address the relationship of CT findings to SUR1 levels in this cohort. This important limitation should be addressed in future prospective studies of this pathway in pediatric patients. A further limitation is the relatively low mean ICP of patients in this sample; none had a mean ICP ≥20 mm Hg. The threshold for raised ICP in pediatric TBI, however, may be lower than 20 mm Hg and/or age dependent.²⁴ In any case, these relatively low values of ICP would be more likely to bias our results toward the null rather than a false positive association.

Conclusion

In this exploratory study using samples from a multi-center RCT, SUR1 expression after pediatric TBI was associated with increased ICP and worse functional outcome. Although trials of pharmacological SUR1 inhibition with glyburide are ongoing in adult TBI and ischemic stroke, its role in pediatric TBI remains to be explored. Given the unmet need for additional therapeutic options in pediatric TBI,⁷ further studies of SUR1 in pediatric patients are warranted.

Funding Information

We would like to thank the NIH NINDS (K23NS101036, to R.M.J.; K23 NS104133, to A.K.A.), the Chuck Noll Foundation (R.M.J. and P.M.K) and the University of Pittsburgh Deans Faculty Advancement Award (to R.M.J.) for generous support.

Author Disclosure Statement

RMJ serves on the scientific advisory board, and is a paid consultant form Biogen Inc. For the remaining authors, no competing financial interests exist.