Genetic Variants Associated With Intraparenchymal Hemorrhage Progression After Traumatic Brain Injury

Abstract

Importance: Intracerebral hemorrhage progression is associated with unfavorable outcome after traumatic brain injury (TBI). No effective treatments are currently available. This secondary injury process reflects an extreme form of vasogenic edema and blood-brain barrier breakdown. The sulfonylurea receptor 1-transient receptor potential melastatin 4 (SUR1-TRPM4) cation channel is a key underlying mechanism. A phase 2 trial of SUR1-TRPM4 inhibition in contusional TBI is ongoing, and a phase 3 trial is being designed. Targeted identification of patients at increased risk for hemorrhage progression may inform prognostication, trial design (including patient selection), and ultimately treatment response.

Objective: To determine whether ABCC8 (SUR1) and TRPM4 genetic variability are associated with intraparenchymal hemorrhage (IPH) progression after severe TBI, based on the putative involvement of the SUR1-TRPM4 channel in this pathophysiology.

Design, setting, and participants: In this genetic association study, DNA was extracted from 416 patients with severe TBI prospectively enrolled from a level I trauma academic medical center from May 9, 2002, to August 8, 2014. Forty ABCC8 and TRPM4 single-nucleotide variants (SNVs) were genotyped (multiplex, unbiased). Data were analyzed from January 7, 2020, to May 3, 2021.

Main outcomes and measures: Primary analyses addressed IPH progression at 6, 24, and 120 hours in patients without acute craniectomy (n = 321). Multivariable regressions and receiver operating characteristic curves assessed SNV and haplotype associations with progression. Spatial modeling and functional predictions were determined using standard software.

Results: Of the 321 patients included in the analysis (mean [SD] age, 37.0 [16.3] years; 247 [76.9%] male), IPH progression occurred in 102. Four ABCC8 SNVs were associated with markedly increased odds of progression (rs2237982 [odds ratio (OR), 2.60-3.80; 95% CI, 1.14-5.90 to 1.80-8.02; P = .02 to P < .001], rs2283261 [OR, 3.37-4.77; 95% CI, 1.07-10.77 to 1.89-12.07; P = .04 to P = .001], rs3819521 [OR, 2.96-3.92; 95% CI, 1.13-7.75 to 1.42-10.87; P = .03 to P = .009], and rs8192695 [OR, 3.06-4.95; 95% CI, 1.02-9.12 to 1.67-14.68]; P = .03-.004). These are brain-specific expression quantitative trait loci (eQTL) associated with increased ABCC8 messenger RNA levels. Regulatory annotations revealed promoter and enhancer marks and strong and/or active brain-tissue transcription, directionally consistent with increased progression. Three SNVs (rs2283261, rs2237982, and rs3819521) in this cohort have been associated with intracranial hypertension. Four TRPM4 SNVs were associated with decreased IPH progression (rs3760666 [OR, 0.40-0.49; 95% CI, 0.19-0.86 to 0.27-0.89; P = .02 to P = .009], rs1477363 [OR, 0.40-0.43; 95% CI, 0.18-0.88 to 0.23-0.81; P = .02 to P = .006], rs10410857 [OR, 0.36-0.41; 95% CI, 0.20-0.67 to 0.20-0.85; P = .02 to P = .001], and rs909010 [OR, 0.27-0.40; 95% CI, 0.12-0.62 to 0.16-0.58; P = .002 to P < .001]). Significant SNVs in both genes cluster downstream, flanking exons encoding the receptor site and SUR1-TRPM4 binding interface. Adding genetic variation to clinical models improved receiver operating characteristic curve performance from 0.6959 to 0.8030 (P = .003).

Conclusions and relevance: In this genetic association study, 8 ABCC8 and TRPM4 SNVs were associated with IPH progression. Spatial clustering, brain-specific eQTL, and regulatory annotations suggest biological plausibility. These findings may have important implications for neurocritical care risk stratification, patient selection, and precision medicine, including an upcoming phase 3 trial design for SUR1-TRPM4 inhibition in severe TBI.

Figure 1.. Spatial Distribution of ABCC8 Single-Nucleotide Variants (SNVs) in Intraparenchymal Hemorrhage (IPH) Progression After Traumatic Brain Injury (TBI)

A, Graph shows the −log₁₀ (P value) of significant cis–expression quantitative trait loci (eQTL) within ABCC8 (y-axis) and their location on the gene (x-axis). Subgraphs show the SNV eQTL P values and chromosomal locations based on different tissue isolates from the Genotype Tissue Expression (GTEx) project (brain-specific eQTL, white area; non–brain eQTL, tan area). ABCC8 is encoded on the reverse strand; exons appear as black bars (exon 39, far left; exon 1, far right). Gray dots indicate locations and corresponding −log₁₀ P values of trans-eQTL. Most brain-specific cis-eQTL are upstream of intron 10, including the 4 ABCC8 SNVs associated with IPH progression (rs2237982, rs2283261, rs3819521, and rs8192695). These SNVs (red lines) are all brain-specific eQTL. No cis-eQTL are reported in the pancreas; all trans-eQTL are downstream in ABCC8. Both cis- and trans-eQTL are distributed evenly throughout the gene in cardiac tissue. B, Only 2.04% of ABCC8 SNVs are brain-specific eQTL, yet all 4 ABCC8 SNVs associated with IPH progression are brain-specific eQTL.

Figure 2.. ABCC8 Single-Nucleotide Variants (SNVs) Associated With Increased Intraparenchymal Hemorrhage (IPH) Progression in Traumatic Brain Injury (TBI)

Violin plots from Genotype Tissue Expression (GTEx) portal of normalized messenger RNA (mRNA) expression levels associated with genotypes of 4 ABCC8 SNVs associated with IPH progression. Shaded regions indicate density distribution of mRNA expression (white bar, median value). The P value for each SNV at each location indicates the value for different expression levels across genotypes for that SNV in the respective tissue location. The m-value denotes the posterior probability that an expression quantitative trait loci (eQTL) effect exists for this tissue based on cross-tissue meta-analyses in the GTEx project (range, 0 to 1; values closer to 1 indicate that the tissue is predicted to have a true eQTL effect). All 4 ABCC8 SNVs are brain-specific eQTL; the 3 intronic SNVs (rs2237982, rs2283261, and rs3819521) are brain-specific eQTL in multiple tissue locations, with m-values close to or equal to 1. In all cases, mRNA expression is higher with SNVs, with a dose-dependent effect.

Figure 3.. Model of Hemorrhage Progression Including ABCC8 and TRPM4 Genotypes vs Standard Clinical Models

Receiver operating characteristic curves for different multivariable models of intraparenchymal hemorrhage (IPH) progression after severe traumatic brain injury (TBI). The simple clinical model consisting of clinical variables significantly associated with IPH progression in this cohort in a backward elimination model (ie, age and initial hemorrhage volume) provides fair discrimination with an area under the curve (AUC) of 0.6959. This is only marginally improved in the full clinical model to 0.7100 by the addition of other clinical characteristics (sex, Glasgow Coma Scale score, Injury Severity Score, partial thromboplastin time, international normalized ratio, and thrombocytopenia). The simple and full clinical models did not perform differently (P = .26). Addition of ABCC8 single-nucleotide variants (SNVs) rs2237982 and rs8192695 and TRPM4 SNVs rs909010 and rs10410857 to the basic model markedly improved model performance with an AUC of 0.8030, approaching excellent discrimination that was superior to both the simple (P = .004) and full (P = .003) models. The remaining significant SNVs were excluded from the model owing to complete overlap in patients with homozygous variants resulting in model overdetermination.

Figure 4.. Three-Dimensional Structure of Sulfonylurea Receptor 1–Transient Receptor Potential Melastatin 4 (SUR1-TRPM4) Pore-Forming Octameric Cation Channel Involved in Hemorrhage Progression After Traumatic Brain Injury (TBI)

A 3-dimensional model of the octameric structure of 4 SUR1 subunits (gray) that assemble with 4 TRPM4 subunits (blue) to create a pore-forming nonselective cation channel. The structures were obtained from the Research Collaboratory for Structural Bioinformatics Protein Data Bank based on work by Li et al on the pancreatic channel SUR1-K_ir6.2 and Autzen et al on the human TRPM4 channel. University of California, San Francisco, Chimera software was used to generate the 3-dimensional structure of SUR1 in this figure without the associated K_ir6.2 channel, which was replaced with TRPM4 in the model. A, Aerial view illustrating the 4 SUR1 subunits binding with the 4 inner TRPM4 subunits. Amino acid sequences encoded by regions of DNA in linkage disequilibrium (LD) with the ABCC8 single-nucleotide variants (SNVs) associated with hemorrhage progression including exon-3 (which contains rs8192695) are highlighted in red. These sequences correspond to the sulfonylurea receptor site, as well as motifs that interface with TRPM4. Amino acid sequences encoded by regions of DNA in LD with TRPM4 SNVs associated with hemorrhage progression are highlighted in black and translate to motifs that interface with SUR1. B, Aerial view rotated 90° around the x-axis to provide a coronal view.