Plasma glial fibrillary acidic protein and neurological diseases: A bidirectional Mendelian randomization study

Abstract

Background

Increased glial fibrillary acidic protein (GFAP) in blood, a biomarker of reactive astrogliosis and astrocytic injury, was observed in a variety of neurological disorders. However, the causal relationship between plasma GFAP and neurological disorders remains unclear.

Objective

We aim to investigate causal association between plasma GFAP levels and neurological disorders using bidirectional Mendelian randomization (MR).

Methods

The genome-wide association studies for neurological disorders, including neurodegenerative diseases, neuroimmune disorders, cerebrovascular diseases, and epilepsy, were collected. Genetic variables associated with plasma GFAP levels were obtained from the UK Biobank Pharma Proteomics Project. Inverse variance weighted or Wald ratio method was used as the main analysis to assess the causal association.

Results

Genetically predicted higher plasma GFAP levels were found to be associated with an increased risk of encephalitis (odds ratio [OR] = 2.52; 95% confidence interval [CI] = 1.67–3.47; p = 1.22 × 10^–5). Furthermore, we found that Alzheimer's disease (β = 0.05; standard error [SE] = 0.01; p = 6.63 × 10^–8), frontotemporal dementia (β = 0.12; SE = 0.01; p = 5.10 × 10^–16), and dementia with Lewy bodies (β = 0.08; SE = 0.02; p = 5.45 × 10^–5) were causally linked to an increase in plasma GFAP levels. Even after controlling for the influence of aging, these associations remained significant.

Conclusions

Our study found that higher plasma GFAP levels may increase the risk of encephalitis, while neurodegenerative dementia may enhance the plasma GFAP levels, supporting the clinical utility of blood GFAP as a reliable biomarker in neurological diseases.

Keywords

aging Alzheimer's disease encephalitis frontotemporal dementia Mendelian randomization plasma GFAP

Introduction

Due to the difficulty in obtaining central nervous system samples, the identification of blood-derived biomarkers for the diagnosis, prognosis, preventive trials, and monitoring of neurological diseases has become an urgent area that requires breakthroughs. Glial fibrillary acidic protein (GFAP) in the blood is increasingly recognized as a biomarker for reactive astrogliosis and astrocytic injury in diverse neurological disorders,^1,2 akin to high-sensitivity troponin serving as a biomarker for myocyte damage in cardiac conditions.³ GFAP is a cytoskeletal protein primarily found in astrocytes. During brain injuries or certain pathological conditions, astrocytes can become activated and undergo changes, resulting in the upregulation and release of GFAP into the surrounding extracellular space and subsequently into the bloodstream. The release process may involve alterations in the glymphatic system and disruption of the blood-cerebrospinal fluid (CSF) barrier.^1,4 Therefore, measuring GFAP levels in the blood may become a valuable tool for assessing the severity and extent of brain damage in conditions such as traumatic brain injury, stroke, and neurodegenerative diseases.

In fact, an increasing body of observational evidence suggests a close association between GFAP and neurological disorders. Compared to healthy controls, elevated GFAP levels in the blood have been observed in various neurological diseases, including acute brain injury (e.g., traumatic brain injury, intracerebral haemorrhage, and ischemic stroke), neurodegenerative diseases [e.g., Alzheimer's disease (AD), frontotemporal dementia (FTD), dementia with Lewy bodies (DLB), and Parkinson's disease (PD)], inflammatory diseases [e.g., multiple sclerosis (MS), neuromyelitis optica spectrum disorder, encephalitis (NMOSD)], and other conditions (e.g., epilepsy and encephalopathy).¹ Importantly, GFAP is closely associated with the pathological changes occurring in some diseases. Evidence suggests that increased plasma GFAP levels may be related to the presence of cortical are amyloid-β (Aβ) deposition in AD patients.^5,6 In addition, elevated levels of GFAP in the blood can be detected early in the disease progression. Recent studies have reported that increased plasma levels of GFAP have been observed in the initial stages of sporadic AD and asymptomatic stages of familial AD.^5,7–10 Plasma GFAP levels are upregulated in individuals who are Aβ positive but cognitively unimpaired.⁷ However, while these observational studies have identified various changes in plasma GFAP levels associated with neurological disorders, some research findings are inconsistent, and the causal relationship or directionality remains unclear. It is worth noting that the early elevation of plasma GFAP levels in disease progression may also imply that GFAP could be a cause rather than a result. Furthermore, evidence suggests that the expression of GFAP by astrocytes gradually increases with age in individuals without neurological disorders,¹¹ which highlights the need to consider the confounding effects of aging when studying the relationship between GFAP and neurological diseases.¹²

Mendelian randomization (MR) is a statistical method that enables the inference of causal relationships between an exposure and an outcome by employing genetic variations.¹³ Single-nucleotide polymorphisms (SNPs) are the most frequently studied genetic variants in MR. These SNPs are randomly allocated to offspring during gamete formation and precede the onset of disease. Consequently, MR is less prone to confounding factors and reverse causation, enhancing its ability to establish causality between variables. Based on these advantages, we employed bidirectional MR analyses in this study to explore the causal relationships and their directions between plasma GFAP levels and neurological disorders including AD, FTD, DLB, PD, ALS, MS, NMOSD, encephalitis, cerebrovascular diseases, and epilepsy. Additionally, we utilized multivariable MR analysis to control for the confounding factor of aging.

Methods

Bidirectional Mendelian randomization design

In this study, we employed a bidirectional MR design to investigate the causal association between plasma GFAP levels and neurological disorders. Publicly available summary-level datasets were obtained from original studies that have received ethical committee approval and informed consent from participants. The validity of the MR study relies on three essential assumptions: (1) a strong correlation between genetic variants and the exposure; (2) no association between genetic variants and confounding factors; and (3) genetic variants solely influence the outcome through the exposure and not through alternative pathways. Figure 1 illustrates the study design.

Figure 1.

Overview of the study design. Parts of figure were drawn by using pictures from SciDraw (https://www.scidraw.io/).

Data sources for GFAP

The publicly available summary statistic for plasma GFAP levels was obtain from the largest plasma proteomic genome-wide association study (GWAS) to date conducted by the UK Biobank Pharma Proteomics Project. This project included a total of 54,219 participants from the UK Biobank.¹⁴ Among these participants, summary statistics for plasma GFAP in the European population were derived from a subset of 32,847 individuals using the Olink Explore 3072 platform.

Data sources for neurological disorders

We included four categories of neurological disorders in our study: neurodegenerative diseases (AD, FTD, DLB, PD, and ALS), neuroimmune disorders (MS, NMOSD, and encephalitis), cerebrovascular diseases (ischemic stroke and its subtypes, as well as intracerebral hemorrhage), and epilepsy (focal and generalized epilepsy; Table 1).

Table 1.

Characteristics of data in this study.

Trait	Consortium	Year	Sample size (cases/controls)	Data source (PMID)
Plasma GFAP	UK Biobank	2023	32847	37794186
Neurodegenerative disease
Alzheimer's disease (AD)	IGAP	2019	21982/41944	30820047
Early onset AD	FinnGen	2021	587/214885	-
Late onset AD	FinnGen	2021	2670/214871	-
Frontotemporal dementia (TDP subtype)	-	2010	515/2509	20154673
Dementia with Lewy bodies	-	2021	2591/6618	33589841
Parkinson's disease	IPDGC	2019	33674/449056	31701892
Amyotrophic lateral sclerosis	AVS	2018	20806/59804	29566793
Neuroimmune diseases
Multiple sclerosis	IMSGC	2019	47429/68374	31604244
NMOSD	-	2018	215/1244	29769526
Encephalitis	FinnGen	2021	398/217308	-
Cerebrovascular diseases
Ischemic stroke (all)	MEGASTROKE	2018	34217/406111	29531354
Large-artery atherosclerosis	MEGASTROKE	2018	4373/406111	29531354
Cardioembolic	MEGASTROKE	2018	7193/406111	29531354
Small-vessel	MEGASTROKE	2018	5386/192662	29531354
Intracranial hemorrhage (Nontraumatic)	FinnGen	2021	2794/203068	-
Epilepsy
Epilepsy (all)	ILAE	2018	15212/29677	30531953
Focal epilepsy	ILAE	2018	9671/29677	30531953
Generalized epilepsy	ILAE	2018	3769/29677	30531953
Aging (PhenoAge)	-	2021	34463	34187551

GFAP: glial fibrillary acidic protein; NMOSD: neuromyelitis optica spectrum disorder; IGAP: International Genomics of Alzheimer's Project; IPDGC: International Parkinson's Disease Genomics Consortium; AVS: ALS Variant Server; IMSGC: International Multiple Sclerosis Genetics Consortium; MEGASTROKE: Multi-ancestry Genome-Wide Association Study of Stroke Consortium; ILAE: International League Against Epilepsy.

Summary statistic for AD were extracted from the International Genomics of Alzheimer's Project (IGAP; 21,982 cases and 41,944 controls).¹⁵ We also obtain the GWAS of AD subtypes from the FinnGen consortium. This included the enrollment of 587 individuals with early-onset AD and 214,885 healthy controls, as well as 2670 individuals with late-onset AD and 214,871 healthy controls. Summary statistic for FTD (TDP subtype) was obtain from an international multicenter study comprising 515 individuals with Frontotemporal lobar degeneration-TAR DNA-binding protein (TDP 43) and 2509 controls of European ancestry.¹⁶ Summary statistic for DLB were derived from another independent multicenter study with a total of 2591 cases and 4027 controls.¹⁷ The International Parkinson's Disease Genomics Consortium (IPDGC) provides access to publicly available GWAS summary statistics for Parkinson's disease, which includes data on of 33,674 PD cases and 449,056 controls European ancestry.¹⁸ In addition, summary statistics for ALS were collected from the ALS Variant Server (AVS), including 20,806 cases and 59,804 controls of European ancestry.¹⁹

The International Multiple Sclerosis Genetics Consortium (IMSGC) conducted a GWAS study on individuals of European ancestry, contributing 14,802 cases and 26,703 controls to investigate genetic associations with multiple sclerosis.²⁰ The genetic association data for NMOSD were obtained from a GWAS meta-analysis including 215 patients and 1244 controls.²¹ We also obtained the summary statistic for encephalitis from the FinnGen consortium, including 398 cases and 217,308 controls of European ancestry.

Summary statistic for ischemic stroke were collected from the Multi-ancestry Genome-Wide Association Study of Stroke (MEGASTROKE) consortium, including 34,217 patients and 406,111 controls of European ancestry.²² We further extracted the GWAS data for three ischemic stroke subtypes in this consortium: large-artery atherosclerotic stroke (4373 cases), cardioembolic stroke (7193 cases), and small-vessel stroke (5386 cases). Additionally, the summary statistic for nontraumatic intracranial hemorrhage was obtained from the FinnGen consortium including 2794cases and 203,068 controls of European ancestry.

Summary statistics for epilepsy and its subtypes were acquired from the International League Against Epilepsy (ILAE) consortium.²³ The dataset included a total of 15,212 cases for all epilepsy and 29,677 controls (approximately 86% of European ancestry), with 3769 cases classified as generalized epilepsy and 9671 cases as focal epilepsy.

Selection of genetic instrumental variables

Figure 1 illustrates the process of selecting genetic IVs to be included in the analyses. Specifically, the selection criteria for extracting the genetic IVs, which are typically SNPs, for plasma GFAP levels and neurological disorders are as follows: (1) genetic IVs were extracted for both plasma GFAP levels and neurological disorders traits at the genomewide significance level (p < 5.0 × 10^–8); (2) genetic IVs underwent clumping at a linkage disequilibrium threshold of R² < 0.001 within ± 10 000 kb distance using the 1000 Genomes European reference panel; (3) weak genetic IVs were excluded from the MR analysis based on the F-statistic [F = (β/se)²] value. In this study, SNPs with F-statistics < 10 were considered weak genetic IVs; (4) palindromic SNPs were excluded from the genetic IVs to ensure that the effects of the SNPs on the exposure were aligned with the same allele effects on the outcome.²⁴

Statistical analyses

To investigate causal effects between the plasma GFAP levels and neurological disorders, we then performed the bidirectional MR analyses. In this study, we utilized Wald's ratio method to estimate the causal effects when there was only one genetic IV available for target exposure. Alternatively, when multiple genetic IVs were available, we employed the inverse variance weighted (IVW) method as the primary approach. If less than three genetic IVs were available, we utilized the fixed-effect IVW method.²⁵ However, if we obtained more than three genetic IVs or encountered significant heterogeneity (Cochran's IVW Q test p < 0.05), we employed the multiplicative random-effect IVW method. By incorporating over-dispersion into the regression model, the multiplicative random-effects model allows for heterogeneity between the causal estimates targeted by the genetic variants. In cases where there were at least three genetic IVs available, we employed various approaches, including MR-Egger, weighted median, and weighted mode to ensure the robustness of the MR results.²⁶ Furthermore, we also applied some sensitivity analyses. Cochran's Q statistics was performed to identify heterogeneity through the IVW and MR-Egger methods.²⁷ To ensure the stability of the causal effect estimates, leave-one-out sensitivity analyses were conducted. Each SNP was systematically excluded from the instrumental variables to examine the presence of outliers and assess the reliability of the results.²⁶ The presence of horizontal pleiotropy was assessed using MR-Egger's intercept, and influential instrumental variables (IVs) affected by pleiotropy were identified using MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) analysis.^28,29 To consider the potential for genetic confounding through aging trait, we further disentangled the relationship between plasma GFAP levels on neurological disorders by adjusting for biological ages (PhenoAge) using the multivariable MR approach.³⁰

A Bonferroni correction was conducted to account for multiple comparisons, with the significance threshold set at 0.05 divided by the number of neurological disorders (0.05/18).²⁴ Only causal associations that passed the Bonferroni correction were considered statistically significant. All statistical analyses were performed using R software (version 4.1.3). The “TwoSampleMR” package (version 0.5.10) was utilized for conducting MR analysis, MR-Egger regression, and Cochran's Q test. The “MRPRESSO” package (version 1.0) was employed for the MR-PRESSO analysis.

Results

After genetic IVs selection, we obtained a total of 15 genetic IVs for the plasma GFAP levels trait and 180 genetic IVs for the 18 neurological disorders traits. No weak genetic IVs were identified as the F-statistics values ranged from 30 to 1269. Supplemental Tables 1 and 2 provides detailed information about the genetic IVs used in the MR analyses for both plasma GFAP levels and f neurological disorders.

Causal effects of GFAP on neurological disorders

The MR results of the causal effects of plasma GFAP levels on neurological disorders were shown in Figure 2 and Supplemental Table 3. Genetically predicted higher plasma GFAP levels were found to be causally associated with an increased risk of encephalitis (odds ratio [OR] = 2.52; 95% confidence interval [CI] = 1.67–3.47; p = 1.22 × 10^–5) based on IVW method. This positive association was consistent with the results obtained from the MR-Egger (OR = 3.44; 95% CI = 1.27–9.28; p = 0.031) and weighted median (OR = 2.46; 95% CI = 1.08–5.61; p = 0.032) approaches. As shown in Table 2 and Supplemental Tables 4–6, the result of Cochran's Q test indicated the absence of heterogeneity (Q-value = 6.07; p = 0.943), and no significant horizontal pleiotropy was found based on the MR-Egger regression intercept analysis (intercept = −0.034; p = 0.454) and the MR-PRESSO global test analysis (p = 0.974). As shown in Figure 3, leave-one-out analysis revealed that no one SNP was responsible for the causal effects of plasma GFAP levels on encephalitis risk. Genetically predicted higher plasma GFAP levels remained significantly associated with an increased risk of encephalitis (p = 0.001) even after adjusting for the effect of genetically predicted aging (Supplemental Table 11). Although a negative link was observed between genetically predicted higher plasma GFAP levels and cardioembolic ischemic stroke risk based on IVW method (OR = 0.77; 95% CI = 0.61–0.99; p = 0.042), it did not pass multiple testing correction and lacked support from other methods.

Figure 2.

MR results of the causal effect of plasma GFAP levels and neurological disorders.

Figure 3.

Scatter plot (A) and Leave-one-out result (B) of the causal effect of plasma GFAP levels on encephalitis risk.

Table 2.

The results of Multivariable MR and sensitivity analyses for the positive causal links between plasma GFAP levels and neurological disorders.

Exposure	Outcome	p (adjusting for aging)	p (Egger intercept)	p_ivw (Cochran's Q)	p (MR-PRESSO)
GFAP	Encephalitis	0.001	0.454	0.943	0.974
AD	GFAP	5.73 × 10^–11	0.117	0.672	0.636
EOAD	GFAP	5.15 × 10^–25	-	-	-
LOAD	GFAP	1.38 × 10^–9	0.154	< 0.001	0.256
FTD (TDP type)	GFAP	5.90 × 10^–18	-	-	-
DLB	GFAP	7.61 × 10^–6	0.174	< 0.001	0.059
CIS	GFAP	0.188	0.757	0.960	0.971

GFAP: glial fibrillary acidic protein; IVW: inverse-variance weighted method; AD: Alzheimer's disease; EOAD: Early onset Alzheimer's disease; LOAD: Late onset Alzheimer's disease; FTD: frontotemporal dementia; DLB: dementia with Lewy bodies; CIS: cardioembolic ischemic stroke.

Causal effects of neurological disorders on GFAP

The MR results of the causal effects of neurological disorders on plasma GFAP levels were shown in Figure 4 and Supplemental Table 7. the IVW method revealed that genetically predicted AD (β = 0.05; standard error [SE] = 0.01; p = 6.63 × 10^–8), late-onset AD (β = 0.06; SE = 0.01; p = 2.05 × 10^–5), DLB (β = 0.08; SE = 0.02; p = 5.45 × 10^–5), and cardioembolic ischemic stroke (β = 0.04; SE = 0.01; p = 6.59 × 10^–8) were found to be causally associated with higher plasma GFAP levels. Those positive association were consistent with the results obtained from the MR-Egger, weighted median and weighted mode approaches (all p < 0.05; Supplemental Table 7). In the sensitivity analyses, no significant horizontal pleiotropy was found in these associations, based on the MR-Egger regression intercept analysis (all p > 0.05) and the MR-PRESSO global test analysis (all p > 0.05; Table 2 and Supplemental Tables 9 and 10). No significant heterogeneity was found in the relationship between AD (Q-value = 14.85; p = 0.672) and cardioembolic ischemic stroke (Q-value = 0.30; p = 0.971) with plasma GFAP levels. However, there was evidence of heterogeneity in the relationship between late-onset AD (Q-value = 27.91; p < 0.001) and DLB (Q-value = 25.65; p < 0.001) with plasma GFAP levels (Table 2 and Supplemental Table 8). As shown in Figure 5, leave-one-out analysis revealed that no one SNP was responsible for the causal effects of AD, late-onset AD, DLB, and cardioembolic ischemic stroke on plasma GFAP levels. Additionally, the Wald ratio method suggested that genetically predicted FTD (β = 0.12; SE = 0.01; p = 5.10 × 10^–16) and early-onset AD (β = 0.06; SE = 0.01; p = 8.53 × 10^–27) were also causally associated with higher plasma GFAP levels. Except for cardioembolic ischemic stroke (p = 0.188), genetically predicted AD (p = 5.73 × 10^–11), early-onset AD (p = 5.15 × 10^–25), late-onset AD (p = 1.38 × 10^–9), FTD (p = 5.90 × 10^–18), and DLB (p = 7.61 × 10^–6) remained significantly associated with plasma GFAP levels even after adjusting for the effect of genetically predicted aging (Table 2 and Supplemental Table 11). Although a positive link was observed between genetically predicted generalized epilepsy and plasma GFAP levels based on IVW method (β = 0.05; SE = 0.02; p = 0.043), it did not pass multiple testing correction and lacked support from other methods.

Figure 4.

MR results of the causal effect of neurological disorders on plasma GFAP levels.

Figure 5.

Scatter plot (A) and Leave-one-out result (B) of the causal effect of Alzheimer's disease, Late onset Alzheimer's disease, Dementia with Lewy bodies, and cardioembolic ischemic stroke on plasma GFAP levels.

Discussion

In this research, we performed the bidirectional MR analyses for the first time to explore the causal relationships between plasma GFAP levels and several neurological disorders. Our results identified that genetically predicted higher plasma GFAP levels were only causally associated with an increased risk of encephalitis. In contrast, multiple neurodegenerative diseases, including AD and its subtypes (early-onset AD and late-onset AD), FTD (TDP subtype), and DLB, were causally linked to higher plasma GFAP levels. These associations remained significant even after adjusting for aging. Additionally, we found that cardioembolic ischemic stroke was positively associated with plasma GFAP levels but influenced by aging. Together, our results suggest that elevated plasma GFAP may exhibit bidirectional roles in neurological diseases, acting as both an upstream causal factor and a downstream biomarker.

Our study suggests that changes in GFAP levels may be a potential causal factor in some neurological diseases. GFAP, as a biomarker for astrocyte injury, may be associated with secondary pathological changes following the onset of the disease. However, it is worth noting that GFAP can also be a primary etiological factor in certain diseases, such as autoimmune GFAP astrocytopathy caused by antibodies targeting GFAP and Alexander disease caused by various dominant heterozygous mutations in the GFAP gene.^1,31 In addition, elevated plasma levels of GFAP detected in the early stages of sporadic AD, the asymptomatic stages of familial AD, and among cognitively unimpaired individuals with positive amyloid-β status suggest that increased GFAP may be involved in early pathological changes and act as an upstream factor at the onset of the disease to promote its development.^5,7–10 Our research reveals that elevated plasma GFAP levels may increase the risk of encephalitis, which suggest that activated or damaged astrocytes make the brain more susceptible to inflammatory attacks or antigen exposure. To gain access to the central nervous system from the bloodstream, pathogens must traverse the blood-brain barrier (BBB), which is formed by endothelial cells and astrocytes.³² Therefore, elevated GFAP levels may also suggest heightened susceptibility to neuroinvasion, potentially acting as an initial factor that promotes the development of encephalitis. Its underlying mechanisms may be associated with astrocyte-mediated neuroinflammation and BBB disruption. Previous studies have supported the association between GFAP levels and encephalitis. Studies have shown elevated GFAP in the cerebrospinal fluid of patients with herpes simplex virus and varicella-zoster virus encephalitis.^33,34 Additionally, increased GFAP levels in the cerebrospinal fluid and serum of patients with tick-borne encephalitis are associated with disease severity.³⁵ Interestingly, autoimmune GFAP disease can occur following infectious encephalitis.³⁶ However, these observational studies cannot establish the directionality of the associations. Further animal research is warranted to explore the differential risk of infection or autoimmune encephalitis based on different states of astrocyte involvement. However, it should be noted that the GWAS data for encephalitis included in this study has a small number of cases, which may lead to unstable statistical results. Therefore, the results regarding the association between GFAP and encephalitis need to be interpreted with caution. Additionally, the lack of information on specific encephalitis subtypes prevented us from exploring the relationship between GFAP levels and encephalitis of different etiologies or pathological types in depth. Future research with larger sample sizes and detailed subtype-resolved datasets is urgently needed to validate our current findings.

In addition to potentially serving as an initiating factor in encephalitis, GFAP may also act as a downstream marker of neurodegenerative diseases. GFAP may serve as a biomarker for reactive astrogliosis and astrocytic injury. GFAP levels in the blood can be used to detect mild damage to the central nervous system, thus serving as a biomarker for disease severity, initial progression, and therapeutic monitoring.¹ In this study, we found a causal relationship between neurodegenerative dementia and elevated plasma GFAP levels. Previous research supports our findings. In a dementia cohort study that did not differentiate dementia types, serum GFAP was found to be associated with long-term risk of incident dementia and dementia-specific mortality.¹² Additionally, elevated serum GFAP levels were found to predict cognitive decline in cognitively unimpaired individuals.³⁷ Among all dementia types, the research on the relationship between GFAP and AD is the most extensive. Studies have shown that blood GFAP levels are significantly higher in AD patients compared to health controls and other neurodegenerative diseases, making it a potential diagnostic marker for differentiation.^38,39 Furthermore, serum GFAP can also differentiate between AD and mild cognitive impairment.⁴⁰ Interestingly, while activated astrocytes are not specific to AD pathophysiology, plasma GFAP was found to be more sensitive than Aβ-PET, CSF and plasma p-tau181, and CSF total tau for early AD diagnosis.⁴¹ Importantly, changes in CSF and plasma GFAP can be observed across the AD continuum.^2,42 Elevated plasma GFAP levels have been observed in initial stages of sporadic AD patients,^5,7,8 asymptomatic carriers in familial AD,^9,10 and early-onset AD cases,⁴³ which consistent with our findings. We found that both late-onset AD and early-onset AD can cause increased plasma GFAP levels. Additionally, research has delved into the relationship between GFAP and AD pathology. Plasma GFAP correlates with brain Aβ burden rather than tau protein and this correlation is stronger than CSF GFAP.⁵ Currently, Aβ deposition is considered an early pathological change in AD and precedes tau protein deposition. These results suggest that astrocytic changes associated with elevated GFAP may be involved in the early pathogenesis of AD. In conclusion, our study genetically supports plasma GFAP as a biomarker for AD that is not influenced by age. However, it is important to note that GFAP is not a specific marker for AD. We found that DLB, another disease with Aβ pathology, also exhibits elevated blood GFAP levels, which is also supported by previous studies. One study included autopsy-confirmed Lewy body spectrum disorders with α-synuclein pathology and found that antemortem plasma GFAP was associated with a higher burden of postmortem Aβ.⁴⁴ Additionally, a study encompassing various neurodegenerative diseases found elevated serum GFAP levels in DLB, in addition to AD, while PD patients did not show such elevation.³⁹ Further research is needed to explore the relationship between GFAP and amyloid deposition.

FTD, second in prevalence to AD, exhibits significant heterogeneity in its pathology, genetics, and clinical symptoms. The research findings on the relationship between GFAP and FTD are not as consistent as those for AD. In FTD cohorts that do not differentiate between underlying mechanisms and clinical phenotypes, elevated blood GFAP levels have been found in patients,⁴² but no changes in serum GFAP were observed in the bvFTD subtype.³⁹ In a recent study that included the FTD spectrum, no blood GFAP alterations were detected.³⁸ However, this study included a mixture of progressive supranuclear palsy and corticobasal syndrome. Some studies suggest that the elevation of GFAP in FTD may be associated with specific mutation types.¹ Both symptomatic and asymptomatic carriers of GRN mutations showed increased plasma GFAP levels.^45,46 Our results indicate that FTD of the TDP subtype may cause elevated plasma GFAP levels, suggesting a potential association between GFAP and specific pathological subtypes. It is noteworthy that the pathological type associated with GRN gene mutations is related to TDP-43 protein.⁴⁷ In addition, a recent post-mortem study found increased GFAP expression in the frontal and temporal cortices of frontotemporal lobar degeneration with tau.⁴⁸ The relationship between GFAP and FTD requires further investigation with larger sample sizes and clarification of different pathological subtypes.

The elevation of plasma GFAP induced by neurodegenerative diseases may involve at least two mechanisms. On the one hand, neurodegeneration-induced astrocyte dysfunction and GFAP release. Neurodegenerative diseases are characterized by progressive neuronal loss and astrocyte activation (reactive astrogliosis), the latter being a compensatory but ultimately detrimental response.⁴⁹ Astrocytes play a critical role in maintaining neuronal homeostasis, but they undergo a series of pathological changes in degenerative states. Chronic neuronal injury triggers astrocyte hypertrophy and upregulation of GFAP expression, reflecting attempts to clear protein aggregates (such as amyloid-β and tau proteins) and mediate neuroinflammation.^50–52 However, prolonged activation leads to astrocyte dysfunction, including impaired glutamate clearance and abnormal synaptic pruning, thereby exacerbating neuronal damage.^53,54 On the other hand, BBB disruption serves as a mechanistic link. Composed of endothelial cells and astrocytic end-feet, the BBB tightly regulates central nervous system homeostasis.⁵⁵ Central nervous system injury triggers astrocytic responses, increasing the expression of GFAP. When the BBB is damaged, GFAP also appears in the periphery.⁵⁶ Neurodegenerative processes can compromise BBB integrity through multiple pathways. In diseases such as AD, astrogliosis and microglial activation release pro-inflammatory cytokines, which disrupt tight junctions between endothelial cells and increase BBB permeability.^57,58 Additionally, studies have shown that plasma GFAP levels in individuals with neurodegenerative diseases are associated with biomarkers of BBB disruption, supporting this mechanistic hypothesis.⁵⁹ Notably, the elevation of plasma GFAP observed in neurodegenerative diseases may not merely represent a marker of injury. GFAP expression is accompanied by astrocyte activation, which acts as a “double-edged sword” in neurological diseases. Some astrocyte subtypes are neurotoxic, while others may promote central nervous system repair.^60,61 Therefore, increased peripheral GFAP levels might also reflect a compensatory adaptation.

Studies have shown that acute stroke, due to sudden BBB disruption and subsequent brain injury, may lead to elevated levels of GFAP in the blood, which is further related to stroke prognosis. Our findings suggest that cardioembolic ischemic stroke is associated with an increase in plasma GFAP levels, but it is influenced by aging. Cardioembolic ischemic stroke, commonly caused by atrial fibrillation, generally presents with a rapid onset and more severe symptoms compared to other types of ischemic stroke. A recent cohort study demonstrated that serum GFAP is helpful in predicting functional outcome after severe acute ischemic stroke.⁶² Interestingly, circulating levels of GFAP were found to be elevated in patients with atrial fibrillation compared to the control group.⁶³ Our study did not find a relationship between GFAP and PD, ALS, MS, and NMOSD. Although some studies support these negative findings,³⁹ we cannot exclude the possibility of false negatives due to certain limitations of this study.

Our study has several advantages. Firstly, we included GWAS data based on a larger sample size and employed MR analysis to comprehensively evaluate the bidirectional relationship between a wide range of neurological disorders and plasma GFAP levels. This is the first study to simultaneously investigate the relationship between multiple different types of neurological disorders and GFAP. Secondly, we conducted multiple sensitivity analyses to validate the stability of the results. Importantly, we accounted for the influence of age on GFAP by conducting multivariable MR analysis to adjust for confounding factors and further ensure the robustness of the findings. However, there are some limitations that need to be considered when interpreting our study results. Firstly, an important limitation of this study is the inclusion of GWAS data on encephalitis, which had a small number of patients, and the specific types of encephalitis were unknown. Therefore, the relationship between GFAP and encephalitis needs to be interpreted with caution, as it cannot be determined whether autoimmune GFAP disorders were included. Secondly, due to the differences in sample sizes and source cohorts for GWAS data of each disease, as well as the lack of individual-level data, we cannot infer the differences in the impact of different diseases on elevated GFAP levels from our results. Thirdly, since GWAS data for other pathological subtypes of FTD (such as the more common tau pathology) is currently unavailable, our study cannot further explore the relationship and differences between plasma GFAP levels and different pathological subtypes of FTD. Fourthly, the GWAS data we employed in this study is mainly derived from European populations, limiting the generalizability of our study results to other racial/ethnic populations. Finally, the plasma GFAP levels may be influenced by various unknown factors, including secretion from glial cells in other peripheral tissues, and plasma GFAP may not fully reflect the function of astrocytes.¹ Therefore, our results should be interpreted with caution.

Conclusions

In conclusion, our MR research reveals that higher plasma GFAP levels may increase the risk of encephalitis, while neurodegenerative dementia including both early-onset and late-onset AD, FTD (TDP subtype), and DLB may enhance the plasma GFAP levels, independent of the effects of aging. These findings provide genetic insights and highlight the clinical value of monitoring blood GFAP for the diagnosis and management of neurological disorders.

Supplemental Material

sj-xlsx-1-alz-10.1177_13872877251362496 - Supplemental material for Plasma glial fibrillary acidic protein and neurological diseases: A bidirectional Mendelian randomization study

Supplemental material, sj-xlsx-1-alz-10.1177_13872877251362496 for Plasma glial fibrillary acidic protein and neurological diseases: A bidirectional Mendelian randomization study by Deming Jiang, Ailing Yue, Haitian Nan, Jiahui Hou, Min Chu, Yihao Wang and Liyong Wu in Journal of Alzheimer's Disease

Footnotes

Acknowledgements

The authors thank the UK Biobank, FinnGen, IPDGC, AVS, IMSGC, MEGASTROKE, and ILAE consortium for providing GWAS summary data.

ORCID iDs

Deming Jiang

Min Chu

Liyong Wu

Ethical considerations

No individual-level data was utilized in this study. Therefore, no new ethical review board approval was required. The research utilized published studies and consortia that offer publicly accessible summary statistics. The original studies included in this research have obtained approval from their respective ethical review boards, and participants have provided informed consent.

Consent to participate

This study only used publicly available summary statistics from published genome-wide association studies. No individual-level data were involved, and no additional informed consent is needed in this study.

Author contribution(s)

Deming Jiang: Conceptualization; Data curation; Methodology; Visualization; Writing – original draft.

Ailing Yue: Data curation; Writing – review & editing.

Haitian Nan: Writing – review & editing.

Jiahui Hou: Writing – review & editing.

Min Chu: Writing – review & editing.

Yihao Wang: Writing – review & editing.

Liyong Wu: Conceptualization; Writing – review & editing.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (82271464).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

The GWAS data used in the analysis are publicly available (Table 1). The GWAS data for plasma GFAP was obtain from the UK Biobank Pharma Proteomics Project (https://metabolomips.org/ukbbpgwas/). The GWAS data for AD, FTD (TDP subtype), DLB, PD, ALS, MS, Encephalitis, cerebrovascular diseases, epilepsy, and PhenoAge were collected from OpenGWAS (https://gwas.mrcieu.ac.uk/). The GWAS data for NMOSD was obtain from GWAS Catalog (; GCST005964).

Supplemental material

Supplemental material for this article is available online.

References

Abdelhak

Foschi

Abu-Rumeileh

, et al. Blood GFAP as an emerging biomarker in brain and spinal cord disorders. Nat Rev Neurol 2022; 18: 158–172.

Pelkmans

Shekari

Brugulat-Serrat

, et al. Astrocyte biomarkers GFAP and YKL-40 mediate early Alzheimer’s disease progression. Alzheimers Dement 2024; 20: 483–493.

Sandoval

Apple

Mahler

, et al. High-sensitivity cardiac troponin and the 2021 AHA/ACC/ASE/CHEST/SAEM/SCCT/SCMR guidelines for the evaluation and diagnosis of acute chest pain. Circulation 2022; 146: 569–581.

Escartin

Galea

Lakatos

, et al. Reactive astrocyte nomenclature, definitions, and future directions. Nat Neurosci 2021; 24: 312–325.

Pereira

Janelidze

Smith

, et al. Plasma GFAP is an early marker of amyloid-β but not tau pathology in Alzheimer’s disease. Brain 2021; 144: 3505–3516.

Asken

Elahi

La Joie

, et al. Plasma glial fibrillary acidic protein levels differ along the spectra of amyloid burden and clinical disease stage. J Alzheimers Dis 2021; 80: 471–474.

Chatterjee

Pedrini

Stoops

, et al. Plasma glial fibrillary acidic protein is elevated in cognitively normal older adults at risk of Alzheimer’s disease. Transl Psychiatry 2021; 11: 27.

Benedet

Milà-Alomà

Vrillon

, et al. Differences between plasma and cerebrospinal fluid glial fibrillary acidic protein levels across the Alzheimer disease continuum. JAMA Neurol 2021; 78: 1471–1483.

O’Connor

Abel

Benedet

, et al. Plasma GFAP in presymptomatic and symptomatic familial Alzheimer’s disease: a longitudinal cohort study. J Neurol Neurosurg Psychiatry 2023; 94: 90–92.

10.

Chatterjee

Vermunt

Gordon

, et al. Plasma glial fibrillary acidic protein in autosomal dominant Alzheimer’s disease: associations with Aβ-PET, neurodegeneration, and cognition. Alzheimers Dement 2023; 19: 2790–2804.

11.

Nichols

Day

Laping

, et al. GFAP mRNA increases with age in rat and human brain. Neurobiol Aging 1993; 14: 421–429.

12.

Cronjé

Liu

Odden

, et al. Serum NfL and GFAP are associated with incident dementia and dementia mortality in older adults: the cardiovascular health study. Alzheimers Dement 2023; 19: 5672–5680.

13.

Emdin

Khera

Kathiresan

. Mendelian randomization. JAMA 2017; 318: 1925–1926.

14.

Sun

Chiou

Traylor

, et al. Plasma proteomic associations with genetics and health in the UK biobank. Nature 2023; 622: 329–338.

15.

Kunkle

Grenier-Boley

Sims

, et al. Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet 2019; 51: 414–430.

16.

Van Deerlin

Sleiman

PMA

Martinez-Lage

, et al. Common variants at 7p21 are associated with frontotemporal lobar degeneration with TDP-43 inclusions. Nat Genet 2010; 42: 234–239.

17.

Chia

Sabir

Bandres-Ciga

, et al. Genome sequencing analysis identifies new loci associated with Lewy body dementia and provides insights into its genetic architecture. Nat Genet 2021; 53: 294–303.

18.

Nalls

Blauwendraat

Vallerga

, et al. Identification of novel risk loci, causal insights, and heritable risk for Parkinson’s disease: a meta-analysis of genome-wide association studies. Lancet Neurol 2019; 18: 1091–1102.

19.

Nicolas

Kenna

Renton

, et al. Genome-wide analyses identify KIF5A as a novel ALS gene. Neuron 2018; 97: 1268–1283.e6.

20.

International Multiple Sclerosis Genetics Consortium. Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science 2019; 365: eaav7188.

21.

Estrada

Whelan

Zhao

, et al. A whole-genome sequence study identifies genetic risk factors for neuromyelitis optica. Nat Commun 2018; 9: 1929.

22.

Malik

Chauhan

Traylor

, et al. Multiancestry genome-wide association study of 520,000 subjects identifies 32 loci associated with stroke and stroke subtypes. Nat Genet 2018; 50: 524–537.

23.

International League Against Epilepsy Consortium on Complex Epilepsies. Genome-wide mega-analysis identifies 16 loci and highlights diverse biological mechanisms in the common epilepsies. Nat Commun 2018; 9: 5269.

24.

Wang

, et al. Causal association of iron status with functional outcome after ischemic stroke. Stroke 2024; 55: 423–431.

25.

Chen

Peters

Prins

, et al. Systematic Mendelian randomization using the human plasma proteome to discover potential therapeutic targets for stroke. Nat Commun 2022; 13: 6143.

26.

Dong

M-H

Zhou

L-Q

Tang

, et al. CSF sTREM2 in neurological diseases: a two-sample Mendelian randomization study. J Neuroinflammation 2022; 19: 79.

27.

Burgess

Bowden

Fall

, et al. Sensitivity analyses for robust causal inference from Mendelian randomization analyses with multiple genetic variants. Epidemiology 2017; 28: 30–42.

28.

Bowden

Davey Smith

Burgess

. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015; 44: 512–525.

29.

Verbanck

Chen

C-Y

Neale

, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018; 50: 693–698.

30.

McCartney

Min

Richmond

, et al. Genome-wide association studies identify 137 genetic loci for DNA methylation biomarkers of aging. Genome Biol 2021; 22: 194.

31.

Gravier-Dumonceau

Ameli

Rogemond

, et al. Glial fibrillary acidic protein autoimmunity: a French cohort study. Neurology 2022; 98: e653–e668.

32.

Zuza

Barros

HLS

de Mattos Silva Oliveira

, et al. Astrocyte response to St. Louis encephalitis virus. Virus Res 2016; 217: 92–100.

33.

Studahl

Rosengren

Günther

, et al. Difference in pathogenesis between herpes simplex virus type 1 encephalitis and tick-borne encephalitis demonstrated by means of cerebrospinal fluid markers of glial and neuronal destruction. J Neurol 2000; 247: 636–642.

34.

Grahn

Hagberg

Nilsson

, et al. Cerebrospinal fluid biomarkers in patients with varicella-zoster virus CNS infections. J Neurol 2013; 260: 1813–1821.

35.

Veje

Griška

Pakalnienė

, et al. Serum and cerebrospinal fluid brain damage markers neurofilament light and glial fibrillary acidic protein correlate with tick-borne encephalitis disease severity-a multicentre study on Lithuanian and Swedish patients. Eur J Neurol 2023; 30: 3182–3189.

36.

Cheng

Huang

Yang

, et al. Autoimmune GFAP astrocytopathy after viral encephalitis: a case report of bimodal overlapping encephalitis. Front Immunol 2023; 14: 1258048.

37.

Verberk

IMW

Laarhuis

van den Bosch

, et al. Serum markers glial fibrillary acidic protein and neurofilament light for prognosis and monitoring in cognitively normal older people: a prospective memory clinic-based cohort study. Lancet Healthy Longev 2021; 2: e87–e95.

38.

Sanchez

Wilkinson

Coughlan

, et al. Association of plasma biomarkers with cognition, cognitive decline, and daily function across and within neurodegenerative diseases: results from the Ontario Neurodegenerative Disease Research Initiative. Alzheimers Dement 2024; 20: 1753–1770.

39.

Oeckl

Halbgebauer

Anderl-Straub

, et al. Glial fibrillary acidic protein in serum is increased in Alzheimer’s disease and correlates with cognitive impairment. J Alzheimers Dis 2019; 67: 481–488.

40.

Fang

Dai

, et al. Evaluation of serum neurofilament light chain and glial fibrillary acidic protein in the diagnosis of Alzheimer’s disease. Front Neurol 2024; 15: 1320653.

41.

Guo

Shen

X-N

Wang

H-F

, et al. The dynamics of plasma biomarkers across the Alzheimer’s continuum. Alzheimers Res Ther 2023; 15: 31.

42.

Wojdała

Bellomo

Gaetani

, et al. Trajectories of CSF and plasma biomarkers across Alzheimer’s disease continuum: disease staging by NF-L, p-tau181, and GFAP. Neurobiol Dis 2023; 189: 106356.

43.

Cheng

Wang

, et al. High burdens of phosphorylated tau protein and distinct precuneus atrophy in sporadic early-onset Alzheimer’s disease. Sci Bull (Beijing) 2023; 68: 2817–2826.

44.

Cousins

KAQ

Irwin

Chen-Plotkin

, et al. Plasma GFAP associates with secondary Alzheimer’s pathology in Lewy body disease. Ann Clin Transl Neurol 2023; 10: 802–813.

45.

Hsiao-Nakamoto

Chiu

C-L

VandeVrede

, et al. Alterations in lysosomal, glial and neurodegenerative biomarkers in patients with sporadic and genetic forms of frontotemporal dementia. bioRxiv 2024; DOI:10.1101/2024.02.09.579529 [Preprint]. Posted February 12, 2024.

46.

Heller

Foiani

Moore

, et al. Plasma glial fibrillary acidic protein is raised in progranulin-associated frontotemporal dementia. J Neurol Neurosurg Psychiatry 2020; 91: 263–270.

47.

Shao

Todd

, et al. Microglial lysosome dysfunction contributes to white matter pathology and TDP-43 proteinopathy in GRN-associated FTD. Cell Rep 2021; 36: 109581.

48.

Hartnell

Woodhouse

Jasper

, et al. Glial reactivity and T cell infiltration in frontotemporal lobar degeneration with tau pathology. Brain 2024; 147: 590–606.

49.

Jiwaji

Hardingham

. Good, bad, and neglectful: astrocyte changes in neurodegenerative disease. Free Radic Biol Med 2022; 182: 93–99.

50.

Muñoz-Ballester

Robel

. Astrocyte-mediated mechanisms contribute to traumatic brain injury pathology. WIRES Mech Dis 2023; 15: e1622.

51.

Zhang

Peng

, et al. GFAP Expression in injured astrocytes in rats. Exp Ther Med 2017; 14: 1905–1908.

52.

De Bastiani

Bellaver

Brum

, et al. Hippocampal GFAP-positive astrocyte responses to amyloid and tau pathologies. Brain Behav Immun 2023; 110: 175–184.

53.

Scofield

Kalivas

. Astrocytic dysfunction and addiction: consequences of impaired glutamate homeostasis. Neuroscientist 2014; 20: 610–622.

54.

Gomez-Arboledas

Fonseca

Kramar

, et al. C5ar1 signaling promotes region- and age-dependent synaptic pruning in models of Alzheimer’s disease. Alzheimers Dement 2024; 20: 2173–2190.

55.

Ballabh

Braun

Nedergaard

. The blood-brain barrier: an overview: structure, regulation, and clinical implications. Neurobiol Dis 2004; 16: 1–13.

56.

Ayala-Guerrero

García-delaTorre

Sánchez-García

, et al. Serum levels of glial fibrillary acidic protein association with cognitive impairment and type 2 diabetes. Arch Med Res 2022; 53: 501–507.

57.

Yue

Hoi

MPM

. Emerging roles of astrocytes in blood-brain barrier disruption upon amyloid-beta insults in Alzheimer’s disease. Neural Regen Res 2023; 18: 1890–1902.

58.

Kaur

Sharma

Deshmukh

. Activation of microglia and astrocytes: a roadway to neuroinflammation and Alzheimer’s disease. Inflammopharmacology 2019; 27: 663–677.

59.

Cicognola

Mattsson-Carlgren

van Westen

, et al. Associations of CSF PDGFRβ with aging, blood-brain barrier damage, neuroinflammation, and Alzheimer disease pathologic changes. Neurology 2023; 101: e30–e39.

60.

Liddelow

Barres

. Reactive astrocytes: production, function, and therapeutic potential. Immunity 2017; 46: 957–967.

61.

Chen

Zhang

, et al. An update on reactive astrocytes in chronic pain. J Neuroinflammation 2019; 16: 140.

62.

Vollmuth

Fiessler

Montellano

, et al. Incremental value of serum neurofilament light chain and glial fibrillary acidic protein as blood-based biomarkers for predicting functional outcome in severe acute ischemic stroke. Eur Stroke J 2024; 9: 751–762.

63.

Galenko

Jacobs

Knight

, et al. Circulating levels of biomarkers of cerebral injury in patients with atrial fibrillation. Am J Cardiol 2019; 124: 1697–1700.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.06 MB