SLAMF1 as a Risk Biomarker for Non-Traumatic SAH: Evidence from a Multi-Omics Study

Abstract

Background

Subarachnoid hemorrhage (SAH) following aneurysmal rupture remains a devastating cerebrovascular event with limited predictive biomarkers. Accurate prediction of aneurysm rupture risk remains a clinical priority, as it could improve risk prediction and reveal potential therapeutic targets. Leveraging UK Biobank proteomic data, we aimed to identify protein markers associated with SAH risk using observational and genetic analyses.

Methods

We analyzed data from 52,916 participants enrolled in the UK Biobank. The analysis involved 3 steps: 1) Longitudinal cox proportional hazards analyses between normalized circulating levels of 2,923 proteins and incident non-traumatic SAH (aneurysmal or non-aneurysmal) adjusting for age, sex, ancestry, smoking status, hypertension, hyperlipidemia, and diabetes; 2) Proteins identified in step 1 (FDR-adjusted p-values < 0.05) underwent Mendelian Randomization (MR) using cis-protein quantitative trait loci; 3) Cellular expression profile of significant proteins were examined using single-cell transcriptomic from immunophenotypic atlas of human hematopoietic progenitors.

Results

We identified 123 incident SAH cases, mean follow-up was 7.06 years (SD: 3.53), mean age was 59.28 (SD: 7.13) and 62% were females. Signaling Lymphocytic Activation Molecule Family Member 1 (SLAMF1) and Ninjurin 1 (NINJ1) were significantly associated (SLAMF1: HR per SD increase 2.18; 95%CI: 1.49–3.18, adjusted P <0.001; for NINJ1: HR 1.85; 95%CI: 1.43–2.40, adjusted P=0.004). MR confirmed the association for SLAMF1 (Inverse Variance Weighted approach OR 1.73; 95%CI 1.26–2.38), with directionality supported through reverse MR (p>0.05). Single-cell transcriptomic analysis demonstrated high SLAMF1 expression in CD4-CTM, CD4-activated, and CD4-naive cells, indicating a possible immunological role in SAH pathophysiology.

Conclusions

Our combined analytical approach identified SLAMF1 as a protein associated with increased SAH risk. SLAMF1, a receptor involved in modulating innate and adaptive immune responses, has been implicated in inflammatory and autoimmune diseases. SLAMF1 and related proteins represent promising biomarkers for SAH risk, potentially enhancing risk stratification, guiding preventive strategies, and informing future therapeutic development. Further research is necessary to explore its mechanistic role SAH development.

Graphical Abstract

Introduction

Non-traumatic subarachnoid hemorrhage (SAH) is a severe cerebrovascular event with high morbidity and mortality, yet predictive biomarkers for risk stratification remain limited.^1,2

Despite its clinical significance, the underlying biological mechanisms contributing to aneurysm rupture are not fully understood, and reliable tools for identifying individuals at high risk are lacking. This uncertainty is further highlighted by the “brain aneurysm paradox”, wherein most ruptured aneurysms are small, yet the majority of small unruptured aneurysms have a low risk of rupture.³ Such discrepancies underscore the limitations of current predictive models and the urgent need for improved biomarkers, which could not only enhance risk stratification but also offer new targets for therapeutic intervention.⁴

Plasma proteins play essential roles in vascular integrity and inflammation, processes that are critical in aneurysm formation and rupture.⁵ Recent advances in high-throughput proteomics now enable the simultaneous assessment of thousands of circulating proteins, offering new opportunities to identify biomarkers associated with SAH risk.

Genetic analyses can complement proteomic findings by leveraging protein quantitative trait loci (pQTL) to support causal associations by removing potential epidemiological confounding. Mendelian Randomization (MR), leveraging genetically predicted protein levels, provides an additional layer of evidence for evaluating potential risk biomarkers.⁶ Additionally, single-cell transcriptomic analysis allows for the identification of cell types that express these candidate proteins, providing insight into their potential biological roles in SAH pathophysiology.⁷ Recent transcriptomic and bioinformatic analyses suggest immune and inflammatory pathways may be central to intracranial aneurysm rupture. ^8–10

In this study, we applied a multi-step proteogenomic approach using UK Biobank data to identify circulating proteins associated with SAH risk. By integrating prospective proteomic profiling, genetic analyses, and single-cell transcriptomics, we aim to uncover biologically relevant protein markers that may improve risk prediction and provide mechanistic insights into SAH development.

Methods

Study Population

This study is based on data from the UK Biobank, a prospective cohort of 503,317 community-dwelling adults aged between 39 and 73, recruited between 2006 and 2010 from across the UK. The data is available to qualified researchers by application. Participants were generally healthy volunteers at baseline and provided blood samples, medical history, and consent for genetic and proteomic analyses; processing of samples is described in detail elsewhere.¹¹ For this analysis, we included individuals with available proteomic data, and excluded those without measured protein levels or with a history of non-traumatic SAH at baseline.

Exposure and Outcome Assessment

Proteomic profiling was performed using an Olink platform, quantifying normalized circulating levels of 2,923 proteins. Olink assay technology and analyses are described in detail elsewhere.^11,12 The primary exposure was normalized protein concentrations in baseline plasma samples.

The outcome was defined as an incident non-traumatic SAH event occurring after baseline assessment, identified through ICD-10 code I60 (Nontraumatic Subarachnoid Hemorrhage). This code includes both aneurysmal and non-aneurysmal causes, and we retained the term “non-traumatic SAH” to reflect the diagnostic coding used. Participants were followed from baseline until the earliest of SAH, death, loss to follow-up, or administrative censoring on December 31, 2022.

Statistical Analysis

We applied a Cox proportional hazards model to assess the association between baseline protein levels and incident SAH, adjusting for age, sex, ancestry, smoking status, hypertension, hyperlipidemia and diabetes. False discovery rate (FDR) adjustment was applied to account for multiple comparisons.

Proteomic profiling was performed on EDTA-plasma samples using Olink proximity extension assays, enabling the relative quantification of 2,923 unique proteins across four multiplex panels (inflammation, oncology, cardiometabolic, and neurology). Protein expression values were reported as Normalized Protein Expression, calculated through log2 transformation and normalization against internal and plate controls to adjust for technical variation. Assays and samples failing quality control thresholds were flagged and excluded according to Olink’s standard protocols.

Proteins identified in the Cox analysis (FDR-adjusted p-value <0.05) were further tested using Mendelian Randomization (MR). This approach followed previously published methods and instrument selection criteria.¹³

Mendelian Randomization Analysis

To explore whether certain proteins might play a causal role in subarachnoid hemorrhage, we used MR, a method that uses genetic variants as instrumental variables to infer causal relationships.. Because genetic variants are randomly assigned during meiosis, MR helps reduce confounding and supports stronger causal inference than traditional observational methods.¹⁴

We focused on protein quantitative trait loci that are in cis-association (±1 million base pairs) of the protein gene (cis-pQTLs), as these are more likely to directly influence protein levels. Genome-wide significant variants (P < 5×10⁻⁸) were clumped to ensure independence (r² < 0.001). Our primary analysis used Inverse Variance Weighted regression to estimate the effect of each genetically predicted protein level on SAH risk.¹⁵ We also applied two additional methods, MR-Egger and MR-PRESSO, which help account for potential bias from variants that might affect the outcome through other pathways, a phenomenon known as horizontal pleiotropy.

To further clarify directionality of effects, we performed reverse MR—where the exposure and outcome are flipped—to test whether SAH itself might influence protein levels. This combination of methods improves confidence in identifying proteins that may play a true biological role in disease.

Integrating Single-Cell Transcriptomic Data

We leveraged a publicly available immunophenotypic atlas of human hematopoietic progenitors published by Zhang et al., 2024, to assess transcript expression at single-cell resolution.¹⁶ This approach enables the mapping of candidate protein expression to specific cell populations, thereby providing insight into the potential cellular contexts in which these proteins may operate. ¹⁶ We performed differential expression analysis comparing the expression of SLAMF1 in CD4 T cell subsets to all other immune cell types using the Wilcoxon rank-sum test. SLAMF1 was then examined among the significantly upregulated transcripts in CD4 T cells to identify potential cellular sources.

Software

Cox analyses were performed using R 4.2.1. MR analyses were performed in Python using Genal v1.3.2¹⁷.

This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines. The UK Biobank data was accessed using application number 58743.

Results

We identified 123 incident SAH cases among 52,916 participants with available proteomic data and without SAH before baseline (n = 98, Figure S1), with a mean follow-up of 7.06 years (SD 3.53). The mean age at baseline was 59.28 years (SD 7.13), and 62.6% (n = 77) of cases were female (Table S1).

A total of 6 proteins (APOF, CRTAC1, KITLG, NINJ1, RPS10, SLAMF1) were associated with SAH risk in the unadjusted analysis, and 3 proteins (NINJ1, RNF168, SLAMF1) remained significant after adjustment for covariates (Tables S2–S3). Following FDR adjustment, only SLAMF1 (HR 1.97 (95% CI 1.33–2.91, FDR-adjusted P <0.001) and NINJ1 (HR 1.85 (95% CI 1.43–2.40, FDR-adjusted P = 0.004) demonstrated significant associations with SAH risk. (Figure 1) MR analysis provided genetic evidence supporting an association between SLAMF1 and SAH risk, with an Inverse Variance Weighted odds ratio of 1.73 (95% CI 1.26–2.38), indicating that individuals with genetically higher SLAMF1 levels have a 73% increased odds of developing SAH (Table S4, Figure 2). This result was consistent across multiple MR methods, reinforcing the potential causal role of SLAMF1. In contrast, MR analysis did not reveal genetic evidence of an association between NINJ1 and SAH risk. (Table S5) Reverse MR analysis, which tests whether SAH might influence protein levels, did not yield significant results, thereby supporting the directionality of the observed association. (Table S6)

Figure 1.

Volcano plot for the prospective associations of circulating proteins with risk of SAH.

Figure 2.

MR Analysis results.

SLAMF1 was significantly overexpressed in CD4 T cell populations compared to other immune cell types (log2 fold change = 1.14, adj. p<0.001). Specifically, high expression was observed in CD4-central memory, CD4-activated, and CD4-naive T cells. (Figure 3).

Figure 3.

Violin Plot showing SLAMF1 expression in CD4-central memory (CD4-CTM), CD4-activated, and CD4-naive T cells. Data from: An immunophenotype-coupled transcriptomic atlas of human hematopoietic progenitors (Zhang et al., 2024).

Discussion

In this proteome-wide survival and genetic analysis, we identified SLAMF1 (Signaling Lymphocytic Activation Molecule Family Member 1), an immune receptor involved in T-cell activation and inflammation, as a potential biomarker for non-traumatic SAH risk. ¹⁸ SLAMF1 is an immune receptor primarily expressed on hematopoietic cells, where it plays an important role in T-cell activation, macrophage function, and inflammatory signaling. It modulates immune responses by facilitating cell-cell interactions, contributing to both innate and adaptive immunity, and has been implicated in various inflammatory and autoimmune conditions.¹⁸ SLAMF1 demonstrated a strong association with SAH in time-to-event analysis after FDR correction, with additional genetic support from MR analysis, suggesting a potential causal role in disease risk. Single-cell transcriptomic analysis revealed high SLAMF1 expression in CD4+ T cell subtypes, highlighting potential immune-mediated mechanisms underlying SAH pathophysiology.

Prior studies have demonstrated immune involvement in the progression of aneurysmal SAH¹⁹. Specifically, it has been shown that CD4+ T lymphocytes and their associated mediators play a key role in the pathogenesis of aneurysmal lesions, not only in intracranial aneurysms but also in abdominal aortic aneurysms. Specifically, CD4+ subsets contribute to tissue degradation and vascular remodeling via localized inflammatory responses, further supporting the role of immune dysregulation in aneurysm pathology.^20–22

While NINJ1 did not show a significant association in Mendelian Randomization analysis, this does not preclude its relevance to subarachnoid hemorrhage pathophysiology. The lack of MR support may reflect a true absence of a causal genetic relationship, or alternatively, a limitation of available genetic instruments, such as weak or insufficiently powered cis-pQTLs. It is also plausible that NINJ1 functions as a downstream effector in inflammatory signaling cascades rather than an upstream causal driver of disease initiation.²³ NINJ1 is known to mediate lytic cell death and macrophage recruitment, key features in vascular inflammation, and is strongly implicated in the development of abdominal aortic aneurysms (AAA). ²⁴ Although direct evidence in cerebral aneurysms is limited, these mechanisms may also be relevant in the progression or rupture of intracranial aneurysms. Further experimental and longitudinal studies are needed to clarify its role across different vascular beds.

This study is limited by the observational nature of proteomic data and the potential for residual confounding despite adjustments. Although the UK Biobank is a large cohort, the number of SAH events (n=123) was limited due to the rarity of the condition. This reduced statistical power, particularly in the context of testing thousands of proteins, may increase the risk of false positives or missed associations. We applied FDR correction to minimize this risk, but findings should be interpreted with caution and require validation in larger or case-enriched cohorts. Additionally, the UK Biobank cohort is predominantly of European ancestry, which may limit generalizability to other populations. SAH incidence, risk factors, and genetic architecture can differ across ancestral groups, and the performance of both proteomic associations and MR analyses may not be consistent in non-European populations. Future studies in more diverse cohorts are needed to validate and extend these findings. Another limitation is the potential inclusion of non-aneurysmal subarachnoid hemorrhage due to reliance on ICD coding; however, the incidence of non-aneurysmal SAH is low and unlikely to substantially impact our findings. Finally, ascertainment was based on hospital inpatient and death registry codes, so non-hospitalized SAH events not resulting in admission or death certification may not have been captured, potentially leading to some underascertainment. Future studies should validate these findings in diverse cohorts and explore mechanistic pathways linking these proteins to SAH development.

Conclusion

Our findings identify SLAMF1 and NINJ1 as potential biomarkers for non-traumatic SAH risk, with genetic support for SLAMF1 only. Transcriptomic evidence further supports an immune-mediated role for SLAMF1. These proteins may be useful for risk stratification and mechanistic studies, guiding future research into immune-driven pathways involved in SAH.

Supplementary Material

Figure S1

Tables S1–S6

Nonstandard abbreviations and nonstandard acronyms

SAH

Subarachnoid Hemorrhage

SLAMF1

Signaling Lymphocytic Activation Molecule Family member 1

MR

Mendelian randomization

NINJ1

Ninjurin 1

FDR

false discovery rate