Glucocorticoid-associated stroke adverse events: Based on safety signals and mechanistic insights via network pharmacology-molecular docking integration

Objectives: To elucidate the risk of stroke adverse events (AEs) associated with glucocorticoids (GCs). Methods: AEs related to GCs and stroke (2004Q1-2025Q2) were collected through the US Food and Drug Administration Adverse Event Reporting System (FAERS), and analyzed using the reporting odds ratio (ROR) and proportional reporting ratio (PRR). Using network pharmacology approaches to elucidate the protein–protein interaction (PPI) network, key targets, and pathways of GCs in stroke, followed by validation via molecular docking. Results: A total of 2004 stroke AEs were associated with GCs. Dexamethasone was involved in 7 positive signals (ROR > 1), prednisolone and methylprednisolone were each involved in 5 positive signals, and hydrocortisone was involved in 1 positive signal. Commonly reported preferred terms (PTs) included brain oedema and intracranial pressure increased. Logistic regression suggested that age and route of administration were potential factors affecting adverse events. The onset time analysis showed that adverse events mainly occurred within 60 days after the start of treatment. GCs were found to have 81 potential targets in stroke. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analyses revealed enrichment in gene expression, plasma membrane, protein kinase activity, and the TNF signaling pathway. Molecular docking results demonstrated that GCs exhibited strong binding affinities to TNF, ADAM17, and ICAM1. Conclusion: The real-world data and network pharmacology indicated that GCs were associated with a significant risk of stroke AEs. Proactively focusing on and monitoring pharmacovigilance is crucial for mitigating severe consequences. 1. Introduction Glucocorticoids (GCs), a class of steroid hormones with potent anti-inflammatory and immunosuppressive effects, have been used clinically for decades. They are widely employed in the management of various autoimmune diseases (such as rheumatoid arthritis and systemic lupus erythematosus), infection-related inflammation (such as chronic obstructive pneumonia and sepsis), and allergic disorders [1, 2] . Statistics show that millions of people worldwide undergo GCs therapy for various diseases each year. However, with the expanding clinical application of GCs and the growing number of treated patients, their adverse effects have emerged as a major limitation to safe GCs use. These adverse effects include osteoporosis, hyperglycemia, hypertension, and major adverse cardiovascular events (MACEs), among others [3] . As one of the most severe MACEs, stroke not only causes permanent neurological deficits such as limb paralysis and cognitive impairment but also significantly increases short-term mortality and long-term disability rates, imposing a substantial burden on patients’ families and the healthcare system [4-6] . Currently, evidence regarding the association between GCs use and increased stroke risk remains controversial. Some studies indicate that the use of GCs significantly increases the incidence of stroke, potentially by promoting water and sodium retention, activating the renin-angiotensin-aldosterone system, and thereby raising blood pressure [7, 8] . Additionally, it may enhance blood viscosity and platelet aggregation, thereby augmenting the risk of thrombosis [9, 10] . Other evidence, however, indicates that some studies have observed no significant correlation between GC use and stroke incidence. Conversely, GCs may offer potential benefits in stroke treatment, such as preventing vasogenic edema, reducing infarct size, and inhibiting lipid peroxidation [11-13] . Some viewpoints suggest that GCs primarily target vascular walls, endothelial cells, and vascular smooth muscle. Thus, cerebral microvascular endothelial cells may strengthen their barrier function via GCs-mediated effects [14] . For instance, a large randomized clinical trial has to some extent confirmed the ability of methylprednisolone to reduce blood-brain barrier disruption. Furthermore, the reduced incidence of symptomatic intracranial hemorrhage further supports methylprednisolone’s potential role in stabilizing the blood-brain barrier (BBB) [15] . Network pharmacology is an effective method to validate the associations between drugs, targets, and diseases, while molecular docking further clarifies the binding affinity between receptors and ligands at the molecular level [16] . The combination of these two approaches plays a significant role in exploring the correlation and mechanism of action between drugs and diseases. As is well known, drug safety is a critical focus in both pharmaceutical development and clinical practice [17] . To our knowledge, there is a paucity of real-world data analyses examining GCs-related stroke AEs, and no studies have specifically investigated differences in stroke AEs among various GCs medications. The US Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) is one of the largest spontaneous reporting system databases currently operated to explore the correlation and risk of glucocorticoid drugs with stroke adverse events. Moreover, the use of network pharmacology and molecular docking techniques to identify key targets related to glucocorticoid-induced stroke adverse events and to explore the corresponding molecular mechanisms provides a reliable scientific basis for ensuring the clinical safety of GCs. 2. Materials and methods 2.1 Data source This study aims to analyze AEs of GCs associated with stroke using the FAERS database. Data extraction spanned from the first quarter of 2004 to the second quarter of 2025, which can be obtained through the following link: https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html. A total of 23,168,942 reports were extracted for this study. After a systematic deduplication strategy based on two criteria was applied to remove duplicate data, 19,344,796 reports were finally retained for subsequent analysis [18] . The deduplication process is illustrated in Figure 1. 2.2 Procedures Reports with GCs approved by the FDA as the primary suspected drugs were collected, including generic and brand names. Data primarily includes patient demographic statistics, adverse events, reporting sources, reporting dates, drug treatment start/end dates, and patient outcomes. All AEs were standardized using Preferred Terms (PTs) derived from the Medical Dictionary for Regulatory Activities (MedDRA 26.1). Detailed classifications of these standardized AEs are provided in Supplementary Table 1. In clinical practice, GCs are frequently administered as free bases or in the form of various salts. Therefore, text mining approaches were employed to systematically categorize drugs that appeared under different names across the collected cases. Additionally, combination formulations containing GCs and other ingredients were excluded. 2.3 Signal Mining To assess the potential association between drug-related stroke AEs and GCs, disproportionality analysis was conducted using two commonly used metrics: the Reporting Odds Ratio (ROR) and Proportional Reporting Ratio (PRR) techniques. The detailed calculation principles and evaluation criteria for these metrics are presented in Supplementary Table 2 [19] . To investigate the potential impact of covariates such as age, sex, body weight, and route of drug administration on stroke AEs, a logistic regression analysis was performed. Origin 2024 was used for data visualization. Data were analyzed using Microsoft Excel 2021, R software (version 4.3.2), and SPSS version 24.0. P < 0.05 was considered to indicate statistical significance. 2.4 Network pharmacology and enrichment analysis Potential targets of GCs were screened and collected from the Swiss Target Prediction (http://www.swisstargetprediction.ch/), SEA (https://sea.bkslab.org/), and Uniprot (http://www.uniprot.org) databases. Targets related to the keyword ”stroke” were retrieved from the Genecards (http://www.genecards.org) and PharmGKB (https://www.pharmgkb.org/) databases. The intersection targets of drugs and diseases were obtained using the Microbioinformatics Platform (http://www.bioinformatics.com.cn/), and a Venn diagram was drawn. The STRING database (http://www.string-db.org/) was used to construct the protein-protein interaction (PPI) network of the drugs, which was then imported into Cytoscape 3.9.1 for analysis of functional interactions between proteins. The DAVID database (https://david.ncifcrf.gov/) was used to perform Gene Ontology (GO) functional enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, with the species option set to ”Homo sapiens”, to explore the potential mechanisms underlying GCs-induced stroke adverse events. 2.5 Molecular docking The 2D structures of small molecule ligands were obtained from the PubChem database (http://pubchem.ncbi.nlm.nih.gov/), and their 3D structures were generated using ChemOffice software. Protein targets were screened from the RCSB PDB database (http://www.rcsb.org/), with high-resolution crystal structures selected as the receptors for molecular docking. Pymol 2.6 software was used to preprocess the proteins by removing water molecules and phosphate groups. Autodock 1.5.6 was employed to process the structures of both the proteins and the small molecules, including adding hydrogen atoms to the proteins and performing hydrogen addition and torsion determination for the small molecule ligands to define the docking box coordinates. Molecular docking was carried out using Autodock Vina to explore the interactions between the proteins and ligands. The optimal conformation of the molecular simulation was determined by comparing the docking scores. The results of the molecular docking were visualized using Discovery Studio 2019 and Pymol 2.6 software [20] . 3. Results 3.1 Descriptive analysis A total of 2,004 AEs were identified across different GCs, with the following distribution: Prednisolone (n = 825, 41.17%), Dexamethasone (n = 656, 32.73%), Betamethasone (n = 36, 1.80%), Methylprednisolone (n = 355, 17.71%), Triamcinolone (n = 55, 2.74%), Hydrocortisone (n = 72, 3.59%), and Cortisone (n = 5, 0.25%). The basic characteristics of AEs associated with different GCs are summarized in Table 1. There were differences in gender distribution among different drugs. For example, there were more male patients in Dexamethasone (n = 311, 50.08%) and Triamcinolone (n = 23, 46.00%), but overall, female patients accounted for a higher proportion. The weight distribution showed that, except for Hydrocortisone patients with a weight less than 50 kg (n = 7, 9.86%), the majority of patients weighted 50～100 kg. Based on age distribution, reports of Prednisolone, Dexamethasone, and Betamethasone mainly involved patients aged 65 - 85 years, while reports of the other four GCs mainly involved patients aged 45 - 64 years. Among reports from countries other than the United States, Prednisolone accounted for 75.20%, Dexamethasone for 73.59%, Betamethasone for 67.74%, Methylprednisolone for 72.14%, Triamcinolone for 42.00%, Hydrocortisone for 57.75%, and Cortisone for 20%. In terms of reporter type, except for Triamcinolone (n = 19, 38.00%) where consumer reports accounted for a higher proportion, physician reports accounted for the highest proportion for other GCs. Among patient outcomes, hospitalization and other severe outcomes were reported at a relatively high frequency. Additionally, the number of stroke AEs increased significantly from 2016 onwards. In addition, considering the low number of AEs associated with Betamethasone, Triamcinolone, and Cortisone, these three GCs were excluded from subsequent analyses. 3.2 Disproportionality Upon comprehensive evaluation, GCs showed a significant signal at the System Organ Class (SOC) level for the nervous system. Further analysis was conducted at the Preferred Term (PT) level, as shown in Table 2. The PTs were ranked by the strength of the ROR signal, with a total of 14 PT signals identified. In their respective drug instructions, it is recorded that increased intracranial pressure with papilledema (pseudotumor cerebri) occurs after treatment, while other PTs do not record this. Among them, Dexamethasone was associated with 7 positive signals (ROR > 1), Prednisolone and Methylprednisolone each had 5 positive signals, while Hydrocortisone had only 1 positive signal: Intracranial pressure increased (ROR = 5.93, 95% Cl = 3.68-9.54). Among these positive PTs, Brain oedema (PT: 10048962) was the most common for Dexamethasone (n = 109, 16.62%) and Methylprednisolone (n = 69, 18.87%), while Intracranial pressure increased (PT: 10022773) was the most common for Prednisolone (n = 86, 10.42%) and Hydrocortisone (n = 17, 23.61%). Logistic regression analysis (Table 3) revealed that factors such as age, body weight, and gender were not significant risk factors for stroke-related AEs associated with Dexamethasone, Methylprednisolone, or Hydrocortisone (P>0.05). In contrast, age had a significant impact on stroke AEs related to Prednisolone (P<0.05). Moreover, different routes of administration were found to be potential confounding factors for stroke AEs related to Dexamethasone and Hydrocortisone (P<0.05). Further analysis of detailed administration routes (Table 4) indicated that for Hydrocortisone, intravenous administration (OR=2.12, 95% CI=1.74 – 2.59) and topical administration (OR=0.05, 95% CI=0.02 – 0.16) showed significant differences compared to oral administration (P<0.05), with intravenous administration being a risk factor and topical administration a protective factor. For Dexamethasone, no significant differences were found among different administration routes (P>0.05). 3.3 Time-to-onset (TTO) analysis As shown in Figure 2, we described the time to onset of stroke AEs associated with four GCs. The onset of stroke AEs associated with the four GCs mainly occurred within 0-60 days. However, there was also a noticeable increase in the incidence of stroke adverse events associated with Dexamethasone and Prednisolone after more than 180 days. The median time to onset of Hydrocortisone-related stroke was 15 days, Methylprednisolone-related stroke was 21 days, Prednisolone-related stroke was 49 days, and Dexamethasone-related stroke was 102 days. In addition, the average time to onset of stroke associated with Hydrocortisone, Methylprednisolone, Prednisolone, and Dexamethasone was 30 days, 55 days, 168 days, and 136 days, respectively. 3.4 Network pharmacology analysis Potential targets of GCs (Dexamethasone, Prednisolone, Methylprednisolone, Hydrocortisone) were obtained from the Swiss Target Prediction and SEA databases, yielding a total of 206 potential targets. A Venn diagram was used to merge and de-duplicate the targets of GCs with those related to stroke AEs, resulting in 81 intersection targets (Figure 3A). These intersection targets were imported into the STRING database to construct a PPI network, as shown in Figure 3B. The PPI enrichment network consisted of 15 nodes and 93 edges. Using Cytoscape, centrality analysis and evaluation were performed on all target genes. The criteria for selection included betweenness centrality, closeness centrality, and degree centrality all being greater than or equal to their respective median values. The median values were as follows: betweenness centrality of 67.62, closeness centrality of 0.0073, and degree centrality of 16.24. Based on these criteria, the key core targets identified were TNF, IL-6, PPARG, CASP3, ESR1, MAPK3, EGFR, GSK3B, MTOR, and ICAM1 (Figure 3C). 3.5 GO and KEGG enrichment analysis The intersection targets were imported into the DAVID database for enrichment analysis to investigate the biological functions of the potential targets and their relevance to stroke. The results showed that a total of 406 GO terms were collected, of which 253 were related to biological processes (BP), 48 to cellular components (CC), and 105 to molecular functions (MF) (Basic parameters: minimum overlap was 3, P-value cutoff was 0.01, and minimum enrichment was 1.5). As shown in Figure 3D, the top 10 terms were displayed and sorted by gene count in a bubble chart. The biological processes mainly involved neuronal apoptosis, gene expression, and regulation of the epidermal growth factor and MAPK cascades. The cellular components mainly involved the plasma membrane, cell surface, and receptor complexes. The molecular functions mainly involved protein kinases, serine kinases, and ATP binding. KEGG pathway analysis identified a total of 138 pathways. The top 20 pathways with the most significant p-values were selected for visualization in a bubble chart, as shown in Figure 3E. The x-axis represents the degree of gene enrichment, the size of the bubbles indicates the enrichment count, and the color intensity represents the p-value. The major enriched pathways include Prolactin signaling pathway, EGFR tyrosine kinase inhibitor resistance, Central carbon metabolism in cancer, Glioma, Endocrine resistance, ErbB signaling pathway, Colorectal cancer, Choline metabolism in cancer, HIF-1 signaling pathway, Growth hormone synthesis, secretion and action, TNF signaling pathway, Hepatitis B, Breast cancer, Kaposi sarcoma-associated herpesvirus infection, Proteoglycans in cancer, Chemical carcinogenesis-receptor activation, Lipid and atherosclerosis, Alzheimer disease, Pathways of neurodegeneration-multiple diseases, and Pathways in cancer. Based on the above GO analysis and KEGG pathway results, the association between GCs and stroke mainly involves gene expression, plasma membrane, protein kinases, and the TNF signaling pathway. Further research will be conducted to investigate whether GCs are involved in the occurrence and development of stroke AEs through the TNF signaling pathway. 3.6 Molecular docking analysis Molecular docking was performed using Autodock and Autodock Vina software. Three targets mediating the TNF signaling pathway (TNF, ICAM1, ADAM17) and four GCs were selected to determine their binding potential. Generally, a binding energy less than -5.0 kcal/mol indicates good binding activity, while a binding energy less than -7.0 kcal/mol indicates strong binding activity. The lower the binding energy, the stronger the binding activity and affinity, and the more stable the conformation [21] . The binding scores of TNF with the four GCs ranged from -6.5 to -7.3, ICAM1 with the four GCs ranged from -5.8 to -6.3, and ADAM17 with the four GCs ranged from -6.8 to -7.0 (Figure 4A). Additionally, representative 3D binding models with the highest scores were visualized for TNF and Hydrocortisone (-7.3), TNF and Dexamethasone (-7.0), ADAM17 and Dexamethasone (-7.0), and ADAM17 and Methylprednisolone (-7.0) (Figure 4B - 4E). The molecular docking structures revealed that GCs exhibited significant binding affinity to the three targets. 4. Discussion GCs have a long history of clinical use, and adverse reactions inevitably accompany their application, making drug safety a paramount concern. Notably, the FAERS database is a large-scale repository freely accessible to the public, containing extensive AE reports submitted by healthcare professionals, consumers, and others. Currently, systematic research on GCs-associated stroke AEs remains controversial. This study represents the first pharmacovigilance investigation utilizing real-world data from the FAERS database to assess the potential risk of stroke-related AEs with GCs, employing both ROR and PRR methodologies. Specifically, descriptive analyses indicate that women report AEs at a higher rate than men during GCs use, with middle-aged and older patients being more susceptible to such events. Age has emerged as a key factor influencing stroke incidence [22-24] . In line with this, epidemiological studies reveal that stroke prevalence in women increases with age—an association that is particularly pronounced in the context of hormone therapy. Combined with women’s longer life expectancy, this results in a disproportionately higher representation of women among stroke patients [25, 26] . The vast majority of reports originate from diverse regions and are primarily submitted by healthcare professionals. This distribution reinforces the generalizability and reliability of the dataset used in the present analysis. We analyzed four types of GCs and found that the most frequently reported AE for Dexamethasone and Methylprednisolone was brain oedema, whereas intracranial pressure increased was the most common AE for Prednisolone and Hydrocortisone. Brain oedema is a potentially fatal pathological condition, mainly divided into vasogenic and cytotoxic oedema [27] . In oedematous brain tissue, excessive accumulation of extracellular fluid leads to increased intracranial pressure, so there is a close pathological link between oedema and intracranial pressure. The use of GCs to alleviate brain oedema and increased intracranial pressure caused by brain tumours is an accepted treatment in the clinic [28] . Since GCs can be used for the treatment of oedema, why do they cause adverse reactions? First of all, the FDA instructions clearly state that dexamethasone can be used for brain edema related to primary or metastatic brain tumors, craniotomy, or head injury. However, there is still some controversy regarding whether GCs can be used for brain edema caused by stroke patients [29, 30] . Second, there is no absolute distinction between therapeutic effects and adverse reactions, under certain conditions, therapeutic effects may transition into adverse reactions. Stroke itself can disrupt the BBB disruption in patients, directly damaging cerebral vascular endothelial cells and increasing BBB permeability. GCs administration may further activate endothelial cells in stroke patients, leading to secondary impairment of BBB permeability. This allows large amounts of fluid and protein from the blood to seep into the brain tissue interstitium through the damaged BBB, further exacerbating vasogenic oedema and increasing intracranial pressure, which is consistent with the results of a prospective study [31] . Kleinschnitz et al. proposed that under hypoxic conditions, GC receptors would undergo a certain degree of degradation, impairing their intrinsic function of stabilizing the BBB. This impairment promotes the progression of combined diseases, resulting in edema and neurological deficits [32] . Interestingly, both age and gender differences can have different effects on brain oedema in stroke patients, and the use of hormones in different patients may produce different effects [33] . Logistic regression analysis identified significant differences in stroke-related AEs associated with Dexamethasone and Hydrocortisone among different administration routes. Further comparison of different routes of administration revealed that intravenous administration of Hydrocortisone was a risk factor compared to oral administration, while topical administration was a protective factor compared to oral administration. Consistent with these findings, meta-analyses comparing oral versus intravenous routes for systemic GCs therapy have demonstrated a significant increase in adverse reactions with intravenous administration relative to oral therapy [34] .Furthermore, topical GC application is generally preferred over systemic administration for minimizing adverse effects, as it reduces systemic exposure to GCs [35] . TTO analysis is recognized as a crucial tool for healthcare professionals to predict and identify AEs. Specifically, this study demonstrated that the majority of AEs occurred within 60 days of GCs initiation, suggesting that healthcare workers should closely monitor the detailed conditions of patients in the early stages of GCs therapy. In addition, the median TTO of the four GCs varied, which may be related to their different pharmacokinetic pharmacodynamic profiles. Based on the duration of action in the body, Hydrocortisone is classified as a short-acting GC, Methylprednisolone as an intermediate-acting GC, and Dexamethasone and Prednisolone as long-acting GCs, which is consistent with their time to onset. As the duration of action in the body increases, the time to onset also increases to a certain extent. However, due to the large number of missing or invalid values in the FAERS database, only 12.13% of the collected reports were included in this study, which may introduce some bias in the results. Network pharmacology and molecular docking analyses were conducted to explore the association between GCs and stroke. PPI analysis identified a total of 81 potential targets, with the top 10 core targets being TNF, IL-6, PPARG, CASP3, ESR1, MAPK3, EGFR, GSK3B, MTOR, and ICAM1. Extensive literature review has shown that these core targets are closely related to stroke [36] . Notably, Chen et al. found that the endothelium is a key component of the BBB, and TNF-mediated endothelial necrosis promotes BBB disruption and the development of brain oedema [37] . This also provides a molecular basis for the positive signal of brain oedema observed in our FAERS analysis. The specific mechanisms underlying stroke AEs have not yet been fully elucidated. Combining the results of GO analysis and KEGG pathway enrichment analysis, it was found that the association between GCs and stroke mainly involves gene expression, plasma membrane, protein kinases, and the TNF signaling pathway. This finding is consistent with the mainstream view of the neuroinflammatory cascade reaction commonly seen in stroke [38] . TNF, as a core ligand of the TNF signaling pathway, exerts its biological effects [39] . ADAM17 is a membrane-bound disintegrin metalloproteinase that acts as an upstream activator of TNF-α and is involved in signal transduction [40] . ICAM1, as a downstream effector molecule, is key to the execution of TNF-mediated inflammatory response [41] . ADAM17, TNF, and ICAM1 play different critical regulatory roles in the pathway. Molecular docking results showed that the four GCs exhibited significant binding affinity to the three core targets and were involved in various biological processes and signaling pathways. This study utilized the FAERS database to mine drug safety information related to GCs. However, several limitations of this research should be acknowledged. First, as a spontaneous reporting system, the FAERS database inherently contains missing data, which restricts further analysis. Second, the assessment of causality between GCs and AEs lacks comprehensive information, and only the correlation between the target drugs and AEs can be determined based on the existing data. Finally, this study did not consider the association between different doses of GCs and the occurrence of each PT. In addition, although this study used network pharmacology and molecular docking techniques to explore the potential mechanisms underlying GCs-induced stroke AEs, further experiments are needed to verify the reliability of the signaling pathways. 5. Conclusion The increasing clinical use of GCs has raised considerable concerns regarding medication safety, particularly regarding stroke-related AEs. This study provides new real-world evidence to supplement the safety profile of GCs in relation to stroke. Integrating network pharmacology and molecular docking validation, the ADAM17, ICAM1, and TNF signaling pathways play important roles in the occurrence and development of stroke induced by GCs. Despite the aforementioned limitations, this work still offers meaningful contributions to the rational clinical use of GCs. By leveraging this approach, the present study delivers clinically actionable references to support healthcare providers in assessing stroke-related risks during GC therapy. Author contributions Feifei Chen: Methodology, Writing – original draft, Funding access. Qingqing Ye: Software, Data management. Yu Sun: Conceptualization, Writing – review and editing. Hongbin Xu: Supervise, Project administration, Writing – review and editing. Funding This study was supported by the Natural Science Fund of Ningbo (2022J214). Declaration of interest The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties. Declaration of generative AI use No generative AI technology was used during the writing process. Acknowledgments The FAERS database was acknowledged by the authors since the study was based on the database. Data availability statement The FAERS database utilized in this study is available at https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html.

1. Chen, M., W. Fu, H. Xu, et al., Pathogenic mechanisms of glucocorticoid-induced osteoporosis. Cytokine Growth Factor Rev, 2023. 70:54-66. doi: 10.1016/j.cytogfr.2023.03.002.2. Bruscoli, S., P.G. Puzzovio, M. Zaimi, et al., Glucocorticoids and COVID-19. Pharmacol Res, 2022. 185:106511. doi: 10.1016/j.phrs.2022.106511.3. Vandewalle, J., A. Luypaert, K. De Bosscher, et al., Therapeutic Mechanisms of Glucocorticoids. Trends Endocrinol Metab, 2018. 29(1):42-54. doi: 10.1016/j.tem.2017.10.010.4. Hilkens, N.A., B. Casolla, T.W. Leung, et al., Stroke. Lancet, 2024. 403(10446):2820-2836. doi: 10.1016/s0140-6736(24)00642-1.5. Feigin, V.L., M. Brainin, B. Norrving, et al., World Stroke Organization: Global Stroke Fact Sheet 2025. Int J Stroke, 2025. 20(2):132-144. doi: 10.1177/17474930241308142.6. Ertus, C., L.M. Scailteux, A. Lescoat, et al., Major adverse cardiovascular events in patients with atopic dermatitis treated with oral Janus kinase inhibitors: a systematic review and meta-analysis. Br J Dermatol, 2023. 189(4):368-380. doi: 10.1093/bjd/ljad229.7. Behrend, E.N. and R.J. Kemppainen, Glucocorticoid therapy. Pharmacology, indications, and complications. Vet Clin North Am Small Anim Pract, 1997. 27(2):187-213. doi: 10.1016/s0195-5616(97)50027-1.8. Sundbøll, J., E. Horváth-Puhó, M. Schmidt, et al., Preadmission Use of Glucocorticoids and 30-Day Mortality After Stroke. Stroke, 2016. 47(3):829-35. doi: 10.1161/strokeaha.115.012231.9. Coelho, M.C., C.V. Santos, L. Vieira Neto, et al., Adverse effects of glucocorticoids: coagulopathy. Eur J Endocrinol, 2015. 173(4):M11-21. doi: 10.1530/eje-15-0198.10. Simion, C., E. Campello, E. Bensi, et al., Use of Glucocorticoids and Risk of Venous Thromboembolism: A Narrative Review. Semin Thromb Hemost, 2021. 47(6):654-661. doi: 10.1055/s-0040-1722270.11. de Courten-Myers, G.M., M. Kleinholz, K.R. Wagner, et al., Efficacious experimental stroke treatment with high-dose methylprednisolone. Stroke, 1994. 25(2):487-92; discussion 493. doi: 10.1161/01.str.25.2.487.12. Mujanovic, A., F. Ng, T.R. Meinel, et al., No-reflow phenomenon in stroke patients: A systematic literature review and meta-analysis of clinical data. Int J Stroke, 2024. 19(1):58-67. doi: 10.1177/17474930231180434.13. Altamentova, S., P. Rumajogee, J. Hong, et al., Methylprednisolone Reduces Persistent Post-ischemic Inflammation in a Rat Hypoxia-Ischemia Model of Perinatal Stroke. Transl Stroke Res, 2020. 11(5):1117-1136. doi: 10.1007/s12975-020-00792-2.14. Deng, H.W., W.Y. Mei, Q. Xu, et al., The role of glucocorticoids in increasing cardiovascular risk. Front Cardiovasc Med, 2023. 10:1187100. doi: 10.3389/fcvm.2023.1187100.15. Yang, Q., C. Guo, C. Yue, et al., Methylprednisolone as Adjunct to Endovascular Thrombectomy for Large-Vessel Occlusion Stroke: The MARVEL Randomized Clinical Trial. Jama, 2024. 331(10):840-849. doi: 10.1001/jama.2024.0626.16. Zhang, P., D. Zhang, W. Zhou, et al., Network pharmacology: towards the artificial intelligence-based precision traditional Chinese medicine. Brief Bioinform, 2023. 25(1). doi: 10.1093/bib/bbad518.17. Lasa, J.S., P.A. Olivera, S. Danese, et al., Efficacy and safety of biologics and small molecule drugs for patients with moderate-to-severe ulcerative colitis: a systematic review and network meta-analysis. Lancet Gastroenterol Hepatol, 2022. 7(2):161-170. doi: 10.1016/s2468-1253(21)00377-0.18. Ando, G., K. Taguchi, Y. Enoki, et al., Evaluation of the Expression Time of Ganciclovir-Induced Adverse Events Using JADER and FAERS. Biol Pharm Bull, 2019. 42(11):1799-1804. doi: 10.1248/bpb.b19-00156.19. Evans, S.J., P.C. Waller, and S. Davis, Use of proportional reporting ratios (PRRs) for signal generation from spontaneous adverse drug reaction reports. Pharmacoepidemiol Drug Saf, 2001. 10(6):483-6. doi: 10.1002/pds.677.20. Alexander, N., N. Woetzel, and J. Meiler, bcl::Cluster : A method for clustering biological molecules coupled with visualization in the Pymol Molecular Graphics System. IEEE Int Conf Comput Adv Bio Med Sci, 2011. 2011:13-18. doi: 10.1109/iccabs.2011.5729867.21. Zhang, X., C. Shen, H. Zhang, et al., Advancing Ligand Docking through Deep Learning: Challenges and Prospects in Virtual Screening. Acc Chem Res, 2024. 57(10):1500-1509. doi: 10.1021/acs.accounts.4c00093.22. Ananth, C.V., J.S. Brandt, K.M. Keyes, et al., Epidemiology and trends in stroke mortality in the USA, 1975-2019. Int J Epidemiol, 2023. 52(3):858-866. doi: 10.1093/ije/dyac210.23. Dawson, J. and A. MacDonald, Sex and hypertensive organ damage: stroke. J Hum Hypertens, 2023. 37(8):644-648. doi: 10.1038/s41371-023-00830-0.24. Roy-O’Reilly, M. and L.D. McCullough, Age and Sex Are Critical Factors in Ischemic Stroke Pathology. Endocrinology, 2018. 159(8):3120-3131. doi: 10.1210/en.2018-00465.25. Cherian, L., Women and Ischemic Stroke: Disparities and Outcomes. Neurol Clin, 2023. 41(2):265-281. doi: 10.1016/j.ncl.2022.10.001.26. Zhang, G.Q., J.L. Chen, Y. Luo, et al., Menopausal hormone therapy and women’s health: An umbrella review. PLoS Med, 2021. 18(8):e1003731. doi: 10.1371/journal.pmed.1003731.27. Han, W., Y. Song, M. Rocha, et al., Ischemic brain edema: Emerging cellular mechanisms and therapeutic approaches. Neurobiol Dis, 2023. 178:106029. doi: 10.1016/j.nbd.2023.106029.28. Cenciarini, M., M. Valentino, S. Belia, et al., Dexamethasone in Glioblastoma Multiforme Therapy: Mechanisms and Controversies. Front Mol Neurosci, 2019. 12:65. doi: 10.3389/fnmol.2019.00065.29. Cook, A.M., G. Morgan Jones, G.W.J. Hawryluk, et al., Guidelines for the Acute Treatment of Cerebral Edema in Neurocritical Care Patients. Neurocrit Care, 2020. 32(3):647-666. doi: 10.1007/s12028-020-00959-7.30. Huang, S., L. Liu, X. Tang, et al., Research progress on the role of hormones in ischemic stroke. Front Immunol, 2022. 13:1062977. doi: 10.3389/fimmu.2022.1062977.31. Danielson, M., B. Reinsfelt, A. Westerlind, et al., Effects of methylprednisolone on blood-brain barrier and cerebral inflammation in cardiac surgery-a randomized trial. J Neuroinflammation, 2018. 15(1):283. doi: 10.1186/s12974-018-1318-y.32. Kleinschnitz, C., K. Blecharz, T. Kahles, et al., Glucocorticoid insensitivity at the hypoxic blood-brain barrier can be reversed by inhibition of the proteasome. Stroke, 2011. 42(4):1081-9. doi: 10.1161/strokeaha.110.592238.33. Wan, Y., K.G. Holste, Y. Hua, et al., Brain edema formation and therapy after intracerebral hemorrhage. Neurobiol Dis, 2023. 176:105948. doi: 10.1016/j.nbd.2022.105948.34. Walters, J.A., D.J. Tan, C.J. White, et al., Systemic corticosteroids for acute exacerbations of chronic obstructive pulmonary disease. Cochrane Database Syst Rev, 2014. 2014(9):Cd001288. doi: 10.1002/14651858.CD001288.pub4.35. Reinhart, W.H., [Corticosteroid therapy]. Praxis (Bern 1994), 2005. 94(7):239-43. doi: 10.1024/0369-8394.94.7.239.36. Yao, Y., F. Liu, Z. Gu, et al., Emerging diagnostic markers and therapeutic targets in post-stroke hemorrhagic transformation and brain edema. Front Mol Neurosci, 2023. 16:1286351. doi: 10.3389/fnmol.2023.1286351.37. Chen, A.Q., Z. Fang, X.L. Chen, et al., Microglia-derived TNF-α mediates endothelial necroptosis aggravating blood brain-barrier disruption after ischemic stroke. Cell Death Dis, 2019. 10(7):487. doi: 10.1038/s41419-019-1716-9.38. Alsbrook, D.L., M. Di Napoli, K. Bhatia, et al., Neuroinflammation in Acute Ischemic and Hemorrhagic Stroke. Curr Neurol Neurosci Rep, 2023. 23(8):407-431. doi: 10.1007/s11910-023-01282-2.39. Thougaard, E., P.V. Nielsen, S. Raffaele, et al., Systemic inhibition of soluble TNF significantly changes glial cell populations leading to improved myelin integrity and better functional outcome after experimental stroke. Biomed Pharmacother, 2025. 189:118334. doi: 10.1016/j.biopha.2025.118334.40. Moss, M.L., S.L. Jin, M.E. Milla, et al., Cloning of a disintegrin metalloproteinase that processes precursor tumour-necrosis factor-alpha. Nature, 1997. 385(6618):733-6. doi: 10.1038/385733a0.41. Yan, M., T. Leng, L. Tang, et al., Neuroprotectant androst-3β, 5α, 6β-triol suppresses TNF-α-induced endothelial adhesion molecules expression and neutrophil adhesion to endothelial cells by attenuation of CYLD-NF-κB pathway. Biochem Biophys Res Commun, 2017. 483(2):892-896. doi: 10.1016/j.bbrc.2017.01.030. Figure 1. Flow diagram for the selection of stroke adverse events with glucocorticoids from the FAERS database. Table 1. Characteristics of stroke adverse events correlated with glucocorticoids. | Gender,n(%) | ||||||| | Female | 263 (42.35) | 347 (45.30) | 18 (58.06) | 182 (56.35) | 22 (44.00) | 37 (52.11) | 2 (40.00) | | Male | 311 (50.08) | 332 (43.34) | 9 (29.03) | 124 (38.39) | 23 (46.00) | 26 (36.62) | 2 (40.00) | | Unknown | 47 (7.57) | 87 (11.36) | 4 (12.90) | 17 (5.26) | 5 (10.00) | 8 (11.27) | 1 (20.00) | | Weight(kg),n(%) | ||||||| | ＜50 kg | 22 (3.54) | 16 (2.09) | 0 (0.00) | 12 (3.72) | 1 (2.00) | 7 (9.86) | 0 (0.00) | | ＞100 kg | 9 (1.45) | 8 (1.04) | 1 (3.23) | 9 (2.79) | 4 (8.00) | 5 (7.04) | 0 (0.00) | | 50～100 kg | 105 (16.91) | 104 (13.58) | 7 (22.58) | 57 (17.65) | 8 (16.00) | 5 (7.04) | 3 (60.00) | | Unknown | 485 (78.10) | 638 (83.29) | 23 (74.19) | 245 (75.85) | 37 (74.00) | 54 (76.06) | 2 (40.00) | | Age(years),n(%) | ||||||| | <18 | 60 (9.66) | 72 (9.40) | 0 (0.00) | 40 (12.38) | 6 (12.00) | 7 (9.86) | 0 (0.00) | | 18～44.9 | 87 (14.01) | 113 (14.75) | 5 (16.13) | 78 (24.15) | 8 (16.00) | 19 (26.76) | 1 (20.00) | | 45-64.9 | 152 (24.48) | 196 (25.59) | 5 (16.13) | 99 (30.65) | 18 (36.00) | 18 (25.35) | 3 (60.00) | | 65～84.9 | 196 (31.56) | 228 (29.77) | 8 (25.80) | 67 (20.74) | 8 (16.00) | 17 (23.94) | 1 (20.00) | | >85 | 10 (1.61) | 11 (1.44) | 3 (9.68) | 4 (1.24) | 0 (0.00) | 0 (0.00) | 0 (0.00) | | Unknown | 116 (18.68) | 146 (19.06) | 10 (32.26) | 35 (10.84) | 10 (20.00) | 10 (14.08) | 0 (0.00) | | Reporter Country | ||||||| | US | 149 (23.99) | 180 (23.50) | 9 (29.03) | 80 (24.76) | 29 (58.00) | 29 (40.85) | 4 (80.00) | | No-US | 457 (73.59) | 576 (75.20) | 21 (67.74) | 233 (72.14) | 21 (42.00) | 41 (57.75) | 1 (20.00) | | Unknown | 15 (2.42) | 10 (1.31) | 1 (3.23) | 10 (3.10) | 0 (0.00) | 1 (1.41) | 0 (0.00) | | Reporter type, n(%) | ||||||| | Consumer | 72 (11.59) | 75 (9.79) | 8 (25.80) | 49 (15.17) | 19 (38.00) | 11 (15.49) | 3 (60.00) | | Health Professional | 134 (21.58) | 147 (19.19) | 1 (3.23) | 54 (16.72) | 5 (10.00) | 12 (16.90) | 0 (0.00) | | Pharmacist | 32 (5.15) | 16 (2.09) | 1 (3.23) | 13 (4.02) | 1 (2.00) | 6 (8.45) | 0 (0.00) | | Physician | 235 (37.84) | 247 (32.25) | 9 (29.03) | 106 (32.82) | 16 (32.00) | 22 (30.99) | 1 (20.00) | | Unknown | 148 (23.83) | 281 (36.68) | 12 (38.71) | 101 (31.27) | 9 (18.00) | 20 (28.17) | 1 (20.00) | | Outcomes, n(%) | ||||||| | Death | 147 (23.67) | 226 (29.50) | 2 (6.45) | 91 (28.17) | 4 (8.00) | 14 (19.72) | 1 (20.00) | | Disability | 10 (1.61) | 2 (0.26) | 0 (0) | 1 (0.31) | 5 (10.00) | 0 (0.00) | 0 (0.00) | | Hospitalization | 182 (29.31) | 223 (29.11) | 8 (25.81) | 71 (21.98) | 11 (22.00) | 15 (21.17) | 1 (20.00) | | Life-Threatening | 68 (10.95) | 57 (7.44) | 0 (0.00) | 33 (10.22) | 10 (20.00) | 9 (12.68) | 1 (20.00) | | Other | 214 (34.46) | 258 (33.68) | 21 (67.74) | 127 (39.32) | 20 (40.00) | 33 (46.48) | 2 (40.00) | | Received year, n (%) | ||||||| | 2004-2009 | 47(7.57) | 50(6.53) | 5(16.13) | 44(13.62) | 1(2.00) | 8(11.27) | 1(20.00) | | 2010-2015 | 77(12.40) | 123(16.06) | 6(19.35) | 78(24.15) | 13(26.00) | 5(7.04) | 1(20.00) | | 2016-2020 | 232(37.36) | 322(42.04) | 16(51.61) | 111(34.37) | 15(30.00) | 32(45.07) | 3(60.00) | | 2021-2025Q2 | 265(42.67) | 271(35.38) | 4(12.90) | 90(27.86) | 21(42.00) | 26(36.62) | 0 (0.00) | Table 2. Signal strength of glucocorticoids stroke adverse events at the preferred term level in the FAERS database by ROR and PRR. Table 3. Logistic regression results of stroke adverse events across different factors. | Factors | P-value | Exp（B） | 95% confidence interval of Exp (B) | | | lower limit | upper limit | ||| | Prednisolone | |||| | Age | 0.014 | 1.230 | 1.040 | 1.440 | | Body weight | 0.460 | 0.760 | 0.370 | 1.570 | | Sex | 0.430 | 1.070 | 0.910 | 1.250 | | Route of administration | 0.159 | 0.842 | 0.662 | 1.070 | | Dexamethasone | |||| | Age | 0.352 | 0.920 | 0.780 | 1.090 | | Body weight | 0.513 | 0.810 | 0.420 | 1.540 | | Sex | 0.905 | 0.990 | 0.840 | 1.170 | | Route of administration | 0.002 | 1.332 | 1.107 | 1.602 | | Methylprednisolone | |||| | Age | 0.553 | 0.920 | 0.700 | 1.210 | | Body weight | 0.346 | 1.460 | 0.660 | 3.210 | | Sex | 0.128 | 1.200 | 0.950 | 1.530 | | Route of administration | 0.319 | 0.909 | 0.753 | 1.097 | | Hydrocortisone | |||| | Age | 0.352 | 1.340 | 0.730 | 2.460 | | Body weight | 0.223 | 2.250 | 0.610 | 8.250 | | Sex | 0.614 | 0.860 | 0.470 | 1.560 | | Route of administration | <0.001 | 2.540 | 2.217 | 2.912 | Table 4. Logistic regression results of stroke adverse events across different administration routes. | Factors | P-value | Exp（B） | 95% confidence interval of Exp (B) | | | lower limit | upper limit | ||| | Dexamethasone | |||| | Oral | 1.00(Reference) | ||| | Intravenous | 0.056 | 1.440 | 0.990 | 2.080 | | Topical | 0.966 | 0.000 | 0.000 | - | | Intramuscular | 0.515 | 0.520 | 0.070 | 3.790 | | Intra-articular | 0.140 | 2.990 | 0.700 | 12.740 | | Hydrocortisone | |||| | Oral | 1.00(Reference) | ||| | Intravenous | <0.001 | 2.120 | 1.740 | 2.590 | | Topical | <0.001 | 0.050 | 0.020 | 0.160 | | Intramuscular | 0.944 | 0.000 | 0.000 | - | | Intra-articular | 0.982 | 0.000 | 0.000 | Inf | Figure 2. The onset time of adverse events of the four glucocorticoids. Figure 3. The potential pharmacological mechanisms of glucocorticoids-induced stroke adverse events. (A)Venn diagram of intersectional genes between glucocorticoids and stroke. (B) PPI network of the common targets. (C) PPI network diagram of the top 10 intersecting targets. (D) Gene Ontology (GO) enrichment analysis. (E) Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis. Figure 4. Molecular docking of TNF, ADAM17, and ICAM1 with four types of glucocorticoids.(A) Heatmaps of docking score values for the interactions between glucocorticoids ingredients and TNF, ADAM17, and ICAM1, where a lower score indicates a stronger binding affinity.(B) The representative molecular docking results of TNF and Hydrocortisone.(C) The representative molecular docking results of TNF and Dexamethasone.(D) The representative molecular docking results of ADAM17 and Dexamethasone.(E) The representative molecular docking results of ADAM17 and Methylprednisolone. Information & Authors Information Version history Copyright This work is licensed under a Non Exclusive No Reuse License. Authors Metrics & Citations Metrics Article Usage 118views 125downloads Citations Download citation Feifei Chen, Qingqing Ye, Yu Sun, et al. Glucocorticoid-associated stroke adverse events: Based on safety signals and mechanistic insights via network pharmacology-molecular docking integration. Authorea. 23 October 2025. DOI: https://doi.org/10.22541/au.176121922.23279296/v1 DOI: https://doi.org/10.22541/au.176121922.23279296/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu.