Genetically Determined Levels of Inflammation-Related Proteins and Functional Outcome After Ischemic Stroke: A Mendelian Randomization Study

Article information

J Stroke. 2026;28(2):312-315

Publication date (electronic) : 2026 May 8

doi : https://doi.org/10.5853/jos.2025.05036

Annelie Angerfors¹, Björn Andersson², Sofia Furutjäll¹, Michael Chong³, Christina Jern^1,4,*, Tara M. Stanne^,1,4,*

¹Department of Laboratory Medicine, Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

²Bioinformatics and Data Center, Core Facilities, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden

³Population Health Research Institute, Hamilton Health Sciences and McMaster University, Department of Pathology and Molecular Medicine, Hamilton, ON, Canada

⁴Department of Clinical Genetics and Genomics, Sahlgrenska University Hospital, Region Västra Götaland, Gothenburg, Sweden

Correspondence: Tara M. Stanne Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, Box 440, SE-405 30, Gothenburg, Sweden Tel: +46-31-343-5720 E-mail: tara.stanne@gu.se

*These authors contributed equally as last author.

Received 2025 October 6; Revised 2026 January 27; Accepted 2026 February 4.

Dear Sir:

The inflammatory cascade is a critical driver of both acute tissue damage and long-term recovery following stroke [1]. Therefore, targeting immune ligands or receptors in this cascade has been hypothesized to improve post-stroke outcomes [1]. We recently identified 14 inflammation-related proteins associated with functional outcome 3 months after ischemic stroke [2]. Given that therapeutic targets supported by human genetic evidence are more likely to succeed in clinical trials, we aimed to investigate the potential causal associations between genetically predicted plasma levels of these proteins and post-stroke functional outcome using two-sample Mendelian randomization (MR). Because studies of disease outcomes inherently have a case-only design and often suffer from collider bias, we used a statistical approach to adjust for this bias.

The study design is outlined in Figure 1 and detailed in the Supplementary Methods and Supplementary Tables 1-3. In brief, cis-protein quantitative trait loci from the UK Biobank [3] were used as genetic instruments. Summary statistics from the Genetics of Ischemic Stroke Functional Outcome Genome-Wide Association Study (GWAS) were used to ascertain outcome effects [4]. To address collider bias, we applied Slope-Hunter adjustment using data from the case-control stroke GWAS, GIGASTROKE [5]. All data sources were derived from publicly available summary statistics from individuals of European ancestry. The random-effects inverse-variance weighted method was used. The primary MR analyses were adjusted for age, sex, ancestry, and stroke severity (National Institutes of Health Stroke Scale). Secondary analyses without severity adjustment were also performed. Sensitivity analyses included MR-Egger, MR-Pleiotropy RESidual Sum and Outlier (MR-PRESSO), and MR-Robust Adjusted Profile Score (MR-RAPS). All analyses were conducted in R (version 4.4.1; R Foundation for Statistical Computing, Vienna, Austria). The false discovery rate (FDR) correction was applied; however, because these proteins are highly correlated [2], associations with P<0.05 were considered suggestive.

Figure 1.

Study design and workflow for MR analyzing the causal relation between inflammatory proteins (UK Biobank Pharma Proteomics Project) and functional outcome after ischemic stroke (GISCOME). Genetic instruments were filtered and adjusted for collider bias (GIGASTROKE). Four proteins were ineligible for downstream analysis (IL-6, STAMBP, 4E-BP1, and OSM because of having less than three SNPs and r² <0.005). For the 10 remaining proteins, analyses were performed using both NIHSS-adjusted and unadjusted data, followed by comprehensive sensitivity analyses. GISCOME, Genetics of Ischemic Stroke Functional Outcome; cis-pQTL, cis-protein quantitative trait loci; SNP, single nucleotide polymorphism; NIHSS, National Institutes of Health Stroke Scale; MR, Mendelian randomization; MR-PRESSO, MR-Pleiotropy RESidual Sum and Outlier; MR-RAPS, MR-Robust Adjusted Profile Score.

In the primary MR analysis, significant associations with poststroke functional outcome were observed for genetically determined plasma levels of caspase-8 (CASP-8), hepatocyte growth factor (HGF), and S100 calcium-binding protein A12 (S100A12; EN-RAGE; calgranulin C) (Figure 2A and Supplementary Tables 4-8). A suggestive association was observed with monocyte chemotactic protein-3 (MCP-3; CCL7). Higher genetically determined levels of S100A12 and MCP-3 were associated with an increased risk of unfavorable outcome (odds ratio [OR], 5.67 [95% confidence interval (CI), 1.91–16.86], FDR=0.017; and 2.29 [1.07–4.91], P=0.033, respectively). By contrast, higher genetically determined levels of CASP-8 and HGF were associated with a decreased risk (OR, 0.39 [0.21–0.76]; and 0.09 [0.02–0.47], FDR=0.017 for both).

Figure 2.

Forest plots showing the associations between inflammatory proteins and functional outcome after ischemic stroke using the Slope-Hunter-adjusted random-effects inverse-variance weighted Mendelian randomization (MR) method. (A) Primary analysis using National Institutes of Health Stroke Scale (NIHSS)-adjusted outcome data and (B) secondary analysis without NIHSS adjustment. For each protein, the number of single nucleotide polymorphisms (SNPs) included in the genetic instrument and odds ratios (ORs) with 95% confidence intervals (CIs) are shown. Filled circles indicate associations with P<0.05. The x-axis shows ORs on a logarithmic scale, with OR >1 indicating worse functional outcome and OR <1 indicating better functional outcome.

Directionally concordant results were observed for three proteins in the secondary MR analysis: HGF (OR, 0.14 [0.06–0.37], FDR<0.001), CASP-8 (OR, 0.56 [0.33–0.94], P=0.028), and S100A12 (OR, 3.02 [1.32–6.90], FDR=0.045) (Figure 2B; details provided in Supplementary Tables 9-13). No significant or suggestive associations were observed for any of the remaining proteins.

Our MR analyses indicate that S100A12 and MCP-3 exert a causal detrimental influence on ischemic stroke outcomes, aligning with clinical observations that elevated acute-phase levels are associated with poor outcomes [2]. Importantly, MR reflects a lifelong genetic predisposition to constitutive protein exposure rather than transient post-injury responses. This distinction is particularly relevant for S100A12, which is rapidly induced in microglia and astrocytes following ischemia and functions as an extracellular alarmin/damage-associated molecular pattern during the acute inflammatory response [6]. By contrast, chronic or constitutive upregulation of S100A12 has been implicated in sustained neuroinflammation, synaptic dysfunction, and progressive neuronal loss and has therefore been suggested as a therapeutic target in neuroinflammatory and neurodegenerative diseases [6]. Our findings extend this therapeutic rationale to stroke, suggesting that modulating S100A12, particularly its constitutive activity, may improve long-term outcomes; however, further mechanistic studies are required.

By contrast, our MR findings support a causal protective role for CASP-8 and HGF, despite observational studies linking higher circulating levels to worse outcomes [2]. As both proteins are strongly upregulated during acute stroke [7], these observational associations likely reflect reverse causation, with elevated levels serving as markers of the acute-phase response rather than as mediators of harm. Instead, MR captures constitutive lifelong differences in systemic circulating protein levels and may reflect broader pleiotropic functions. This interpretation aligns with experimental evidence supporting their biologically protective roles in unfavorable post-stroke outcomes. For instance, CASP-8 inhibits necroptosis (a highly proinflammatory form of programmed cell death), microglial activation, and proinflammatory cytokine production [8], whereas HGF enhances angiogenesis and neurogenesis and reduces infarct size when overexpressed in experimental models [9]. Both proteins have been proposed as therapeutic modulators of inflammation, HGF specifically for cerebral ischemia9 and CASP-8 through its role in regulating necroptosis [8]. Our findings suggest that this rationale may plausibly extend to improving post-stroke outcomes. Although our HGF results contrast with those of a previous MR study [10], this inconsistency is likely attributable to methodological differences (Supplementary Results and Supplementary Table 14).

This study followed the Strengthening the Reporting of Observational Studies in Epidemiology–Mendelian Randomization guidelines. Nonetheless it had several limitations. First, we used strict criteria for genetic instrument selection to mitigate horizontal pleiotropy, which resulted in the exclusion of four proteins (IL-6, STAMBP, 4E-BP1, and OSM). Second, genetic instruments were derived from protein levels measured in predominantly healthy individuals, reflecting constitutive systemic biology rather than post-stroke acute-phase changes. Instruments based on post-stroke protein levels could better capture the biology relevant to injury response. As more stroke cohorts undergo broad plasma proteomic profiling, GWAS of acute-phase protein levels will be essential to enable MR studies focused on stroke outcomes. Third, although we adjusted for collider bias, we cannot guarantee that the bias inherent in the case-only design was fully eliminated. Fourth, the analyzed proteins were initially identified in a relatively young ischemic stroke cohort, whereas our outcome datasets span all adult age groups. Fifth, we used only datasets of patients of European ancestry, limiting generalizability to populations with different inflammatory profiles. Finally, as this study was exploratory and lacked replication using an independent dataset, further research is required to provide direct evidence for actionable targets.

In conclusion, we provide evidence supporting the causal detrimental roles of S100A12 and MCP-3 and the protective roles of CASP-8 and HGF in functional outcome after ischemic stroke. Further experimental and clinical studies are warranted to elucidate the precise roles of these proteins in the pathophysiology of acute ischemic stroke, clarify their impact on stroke outcomes, and examine their potential as therapeutic targets.