- Academic Editor
Background: There are current clinical observations that atorvastatin
may promote subdural hematoma resorption. We aimed to assess the causal effects
of lipid-lowering agents 3-hydroxy-3-methylglutaryl coenzyme A reductase
(HMGCR) inhibitors, Proproteinconvertase subtilisin/kexin type
9 (PCSK9) inhibitors and Niemann-Pick C1-like protein 1
(NPC1L1) inhibitors on traumatic subdural hematomas. Methods:
We used genetic instruments to proxy lipid-lowering drug exposure, with genetic
instruments being genetic variants within or near low-density lipoprotein (LDL
cholesterol)-associated drug target genes. These were analyzed by using a
two-sample Mendelian randomization (MR) study. Results: A causal
relationship was found between HMGCR inhibitors and traumatic subdural
hematoma (Inverse variance weighted (
Subdural hemorrhage (SDH), also called traumatic subdural hematoma, often occurs
following traumatic brain injury [1]. SDH may spread rapidly, causing progressive
midline displacement and cerebral hernia, endangering the patient’s life [2]. Additionally, it leads to several complications, such as epilepsy,
hydrocephalus, and chronic subdural hematoma [3, 4, 5]. Subdural hemorrhage occurs
most often after craniocerebral injury, and the incidence is approximately
11%–49% [6]. This places a great burden on individuals, families, and society,
and economic losses can reach
By establishing a rat model of subdural hematoma, Wang et al. [8] found that atorvastatin could promote the absorption of subdural hematoma in the rat model. Atorvastatin also promotes the absorption of chronic subdural hematoma in human patients [9]. Atorvastatin is a 3-hydroxy-3-methylglutaryl coenzyme A reductase reductase (HMGCR) inhibitor, which is the most commonly used class of lipid-lowering drugs. It has several major advantages, such as proven safety, low cost, and multi-potency [10]. Recent studies on atorvastatin have proposed that the mechanism of chronic subdural hemorrhage is stabilizing immature leaky vessel formation in neomembranes [11]. However, the mechanism of traumatic subdural hemorrhage is usually caused by trauma, and Paseban et al. [12] found that the combined use of high-dose aspirin, metformin, captopril, and atorvastatin potentiated their antioxidant effects on the brain and hence could improve cognitive function with their neuroprotective effects on the hippocampus. Therefore, we explored the efficacy of lipid-lowering drugs for acute traumatic subdural hemorrhage. However, current clinical studies are often influenced by confounding factors, which significantly impact the investigation of the causal relationship between exposure and outcome. Furthermore, in these studies, the clinical characteristics of patients using statins and those not using statins were quite different. Because of the retrospective nature of observational studies, causal inference is not allowed; however, Mendelian randomization (MR) analysis can overcome the limitations of traditional observational studies. Therefore, we aimed to explore the relationship between lipid-lowering agents and traumatic subdural hematoma using MR.
MR uses a genetic variation to infer causation between exposure and outcome, thereby assessing whether an observational association is consistent with a causal effect [13]. It is based on the principle of random gamete division and combination in genetic variation to simulate the random assignment process of research objects [14]. Therefore, we used this property to explore the causal relationship between common lipid-lowering drugs and traumatic subdural hematoma.
MR is a statistical method used to estimate the causal effect of an exposure factor on an outcome variable. There are two fundamental principles of Mendelian randomization:
The first is instrumental variable: This is a method derived from econometrics. It is a classical statistical method used to address endogeneity issues and is one of the most commonly used empirical analysis methods in research. The purpose is to identify the causal relationship between the exposure and outcome variables using the exogeneity assumption of instrumental variables (Fig. 1A).
Principle of Mendelian randomization of drug targeting. (A) Working principle of instrumental variables (genetic variation (G); exposure factor (X); outcome (Y)). (B) Schematic diagram of Mendelian laws of inheritance in Mendelian randomization. (C) Working principle diagram of Mendelian randomization of drug targeting. LDL-C, low-density lipoprotein cholesterol; HMGCR, 3-hydroxy-3-methylglutaryl coenzyme A reductase; PCSK9, Proproteinconvertase subtilisin/kexin type 9; NPC1L1, Niemann-Pick C1-like protein 1; SDH, subdural hemorrhage; GWAS, Genome wide association study; TSDH, traumatic subdural hemorrhage.
The second is Mendelian genetics law: Based on the random allocation principle followed by alleles during gamete formation, the genotype can be used as an instrumental variable for the intermediate phenotype under investigation to infer its causal association with the disease status. The estimated effect is not influenced by confounding factors or reverse causation. For this reason, MR is often referred to as a “nature-created randomized controlled trial” [15] (Fig. 1B).
Therefore, MR is a causal inference method based on genetic variation. Its basic principle is to use genetic variants that are randomly allocated in nature to investigate the impact of phenotypes on diseases. In other words, Mendelian randomization is a randomized controlled trial conducted on a population using Mendelian genetics law.
In this study, we introduced the concept of drug targeting. We used single nucleotide polymorphisms (SNPs) associated with low-density lipoprotein cholesterol (LDL-C) reduction at or near the HMGCR, Proproteinconvertase subtilisin/kexin type 9 (PCSK9), and Niemann-Pick C1-like protein 1 (NPC1L1) loci as instrumental variables for statins, PCSK9 inhibitors, and ezetimibe, respectively. The outcome phenotype chosen is the occurrence of traumatic subdural hemorrhage [16] (Fig. 1C).
Sources of LDL-C data: LDL-C Genome wide association study (GWAS) data comprising 2,437,752 individual data points of European ancestry (GWAS ID: ieu-a-300) were collected from an open GWAS database (https://gwas.mrcieu.ac.uk/) [17].
Data sources for traumatic subdural hematoma: The inclusion criteria were ICD-10 code S065 in the Hospital Discharge Registry or Cause of Death Registry and complete GAWS data. Those with incomplete GAWS data were not included in the study. The GWAS data, consisting of the experimental (1682) and control (136,576) groups, were obtained from the FinnGen database (https://www.finngen.fi/fi). All groups were Europeans (GWAS ID: finn-b-ST19_TRAUMAT_SUBDU_HAEMORRHAGE) [18]. The baseline clinical characteristics of the cohort are shown in Table 1.
All | Female | Male | |
Number of individuals | 1682 | 550 | 1132 |
Unadjusted prevalence (%) | 0.78 | 0.45 | 1.2 |
Mean age at first event (years) | 65.23 | 67 | 64.37 |
Case fatality at 5 years (%) | 18.91 | 19.82 | 18.46 |
Median number of events per individual | 1 | 1 | 2 |
Recurrence at 6 months (%) | 43.88 | 35.82 | 47.79 |
According to Mendel’s randomization theory, instrumental variables must satisfy the following three assumptions:
Correlation hypothesis: Instrumental variables must be strongly correlated with exposure factors;
Independence hypothesis: Instrumental variables are independent of confounding factors affecting expose-outcome;
Exclusivity hypothesis: Instrumental variables can only influence the occurrence of results through “exposure” factors but not through other means.
Based on a study by Ference et al. [19], we selected HMGCR, PCSK9, and NPC1L1 inhibitors as the three major common lipid-lowering drugs targeting genes.
To identify potential instrumental variables as substitutes for these drugs, we used the screening method proposed by Huang et al. [20].
(1) SNPs minor allele frequency (MAF)
The instrumental variables that could replace the target drug were obtained by satisfying these four conditions.
The R software (Lucent Technologies, Jasmine Hill, NJ, USA) package
“TwoSampleMR (version 0.5.6, https://github.com/MRCIEU/TwoSampleMR/releases/tag/v0.5.6)” was used for MR data analysis, employing methods
such as inverse-variance weighting, MR Egger, weighted median, simple model, and
weighted model. Inverse-weighted variance analysis was used to determine the
causal relationship between exposure factors and results [21, 22, 23]. Statistical
significance was set at p
MR Egger and inverse variance weighted functions in Cochran Q were used to
assess heterogeneity among the study samples, and p
The horizontal pleiotropy was analyzed using the “mr_pleiotropy test”
function of the R package with the same name, and p
All statistical analyses were performed using the “TwoSampleMR” (version 0.5.6) package in the statistical program R (version 4.1.1) (Lucent Technologies) [24].
Table 2 lists the desired SNPs obtained by satisfying the four major conditions for drug-targeted MR. Seven SNPs (rs10066707, rs10515198, rs12659791, rs12916, rs3804231, rs3857388, and rs72633962) were identified from drug-targeting genes with HMGCR inhibitors (Supplementary Table 1). Twelve SNPs (rs10493176, rs11206510, rs11206514, rs11583974, rs11591147, rs12067569, rs2479394, rs2479409, rs2495495, rs4927193, rs572512 and rs58513) were obtained from drug-targeting genes with PCSK9 inhibitors (Supplementary Table 2). Three SNPs (rs2073547, rs217386, and rs7791240) were obtained from drug-targeting genes with NPC1L1 inhibitors (Supplementary Table 3).
Exposure | Gene ID | Description | Gene region | Genetic variants associated with LDL cholesterol level |
HMGCR inhibitors | 3156 | 3-hydroxy-3-methylglutaryl-CoA reductase | Chr5 75336529-75362116 | (1) SNPs (MAF |
(2) low linkage disequilibrium (r | ||||
(3) associated with LDL cholesterol (p | ||||
(4) located within | ||||
PCSK9 inhibitors | 255738 | proprotein convertase subtilisin/kexin type 9 | Chr1 55039548-5506485 | (1) SNPs (MAF |
(2) low linkage disequilibrium (r | ||||
(3) associated with LDL cholesterol (p | ||||
(4) located within | ||||
NPC1L1 inhibitors | 29881 | NPC1 like intracellular cholesterol transporter 1 | Chr7 44512535-44541330 | (1) SNPs (MAF |
(2) low linkage disequilibrium (r | ||||
(3) associated with LDL cholesterol (p | ||||
(4) located within |
SNPs, single nucleotide polymorphisms; MAF, minor allele frequency; LDL, low-density lipoprotein.
The two-sample analysis yielded the following results: Inverse variance weighted
(
Forest map of Drug target Mendelian randomization results (Expose: HMGCR; Outcome: Traumatic Subdural Hemorrhage). CI, confidence interval; MR, mendelian randomization; OR, odds ratio.
The two-sample analysis yielded the following results: Inverse variance weighted
(
Forest map of Drug target Mendelian randomization results (Expose: PCSK9; Outcome: Traumatic Subdural Hemorrhage).
The two-sample analysis results are as follows: Inverse variance weighted (
Forest map of Drug target Mendelian randomization results (Expose: NPC1L1; Outcome: Traumatic Subdural Hemorrhage).
The Cochran Q test results are as follows:
HMGCR inhibitors: MR-Egger (p = 0.931067) and inverse variance weighted (p = 0.969446) (Table 3).
Heterogene-test | Pleiotropy-test | |||||||
MR Egger | IVW | MR Egger | ||||||
Q | Q_df | Q_pval | Q | Q_df | Q_pval | intercept | SE | pval |
1.337112 | 5 | 0.931067 | 1.339308 | 6 | 0.969446 | –0.004310834 | 0.09198007 | 0.9644335 |
IVW, Inverse variance weighted; SE, Standard Error.
PCSK9 inhibitors: MR-Egger (p = 0.7646204) and inverse variance weighted (p = 0.969446) (Table 4).
Heterogene-test | Pleiotropy-test | |||||||
MR Egger | IVW | MR Egger | ||||||
Q | Q_df | Q_pval | Q | Q_df | Q_pval | intercept | SE | pval |
6.577696 | 10 | 0.7646204 | 10.676316 | 11 | 0.4707652 | 0.04011494 | 0.01981469 | 0.07044835 |
NPC1L1 inhibitors: MR-Egger (p = 0.2809581) and inverse variance weighted (p = 0.4707652) (Table 5).
Heterogene-test | Pleiotropy-test | |||||||
MR Egger | IVW | MR Egger | ||||||
Q | Q_df | Q_pval | Q | Q_df | Q_pval | intercept | SE | pval |
1.162450 | 1 | 0.2809581 | 1.185044 | 2 | 0.5529310 | –0.02829321 | 0.2029396 | 0.9118129 |
These results indicate that our data exhibited no significant heterogeneity and demonstrated good stability, strongly supporting the results of the MR analysis.
The horizontal multi-effect analysis results were insignificant, with p = 0.9644335 for HMGCR inhibitors (Table 3), p = 0.07044835 for PCSK9 inhibitors (Table 4), and p = 0.9118129 for NPC1L1 inhibitors (Table 5). These results indicate the absence of horizontal diversity.
Traumatic SDH is one of the most fatal craniocerebral disorders, with a mortality rate of 40–80%. Despite surgical interventions, patients may experience fatal outcomes or severe disabilities [25]. Because surgery is the primary treatment modality for traumatic SDH, most research on traumatic SDH has mainly focused on the surgical approach and timing. However, with an aging society, the incidence of traumatic cranial injury increases yearly in older adults and is often accompanied by SDH [26]. The importance of pharmacological treatment is gaining prominence, especially among older adults who frequently have comorbidities and medication history (anticoagulants) that pose significant challenges for surgical treatment. Consequently, the importance of pharmacological treatment is becoming apparent.
Statistics reveal that the average out-of-pocket cost per new drug development
is
This study revealed that lipid-lowering drugs targeting HMGCR inhibitors, such as atorvastatin, exhibit a causal and protective effect against traumatic SDH. In contrast, the analysis of lipid-lowering with PCSK9 and NPC1L1 inhibitors as pharmacogenetic targets associated with causality was not supported by the results. These results align with recent studies on brain hemorrhage and lipid metabolism, in which the lipid-lowering effect of statins was not associated with cerebral hemorrhage. However, statins can play a role in improving the prognosis of cerebral hemorrhage [30, 31]. This statement seems contradictory, as Li et al. [31] found that atorvastatin plays an important role in eliminating SDH and improving neurological recovery in a rat model. These mechanisms may improve neurological recovery and anti-inflammatory effects after cerebrovascular disease by reducing the expression of nitric oxide synthase and myeloperoxidase in the brain tissue surrounding the hematoma after cerebral hemorrhage [32].
In this MR study, we used HMGCR-mediated genetic variants associated with LDL-C to represent statin exposure. The MR analysis suggested that HMGCR inhibitors have a protective role in traumatic SDH. The meninges are known to protect and support brain tissue. Especially when the head is subjected to external forces, the meninges can play a protective role, but with the increase of age, the meninges are more and more closely adhered to the skull, and the protection of the meninges is weakened with the increase of skull adhesion. It has been reported that taking atorvastatin can improve tissue adhesion [33]. Therefore, atorvastatin may improve the adhesion between the meninges and the skull, improve the protective function of the meninges and further reduce the incidence of traumatic subdural hemorrhage. Secondly, atorvastatin also specifically improves the effect of neuroinflammation [34], which can reduce the secondary injury of traumatic subdural hemorrhage. This study suggests the need for additional observational, mechanistic, and randomized controlled studies in different populations to examine their potential for treating traumatic SDH. We recommend that patients with traumatic SDH who are currently prescribed or initiating statin therapy should continue their treatment. Statins may be a priority drug in future clinical trials for treating SDH.
This study had several limitations. First, the population in this study was limited to Europeans, lacking ethnical diversity. Our subsequent study will encompass a broader population. Second, the aggregated data used in this study does not adequately support further data analysis. Third, this study did not analyze the complete lipid profile. Finally, this study did not include any in vivo experiments, and the underlying mechanism of statin protection in traumatic subdural hemorrhage was not thoroughly investigated. Therefore, future studies should further explore and elucidate the mechanism behind the potential protective effects of statins in this context.
This MR study suggests a causal relationship between HMGCR inhibition and traumatic SDH. More clinical trials are needed to determine whether statins protect against traumatic subdural hematomas and further studies are needed to explore the underlying mechanisms.
SDH, subdural hemorrhage; MR, Mendelian randomization; LDL-C, low-density lipoprotein cholesterol; HGMCR, 3-hydroxy-3-methylglutaryl coenzyme A reductase; SNP, single nucleotide polymorphism.
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
LFW and KQC designed the research study. HL, HSH and YTC analyzed the data. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.
Not applicable.
Thank you to all the medical staff who contributed to the maintenance of the medical record database (IEU and FinnGen).
This research received no external funding.
The authors declare no conflict of interest.
Publisher’s Note: IMR Press stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.