Transcriptomic features for prediction of endovascular thrombectomy clinical benefit in stroke

Technical Summary: Transcriptomic Features for Prediction of Endovascular Thrombectomy Clinical Benefit in Stroke

Problem Statement
Mechanical thrombectomy (MT) is the standard reperfusion therapy for acute ischemic stroke (AIS) caused by large-vessel occlusion (LVO). However, there is marked interindividual heterogeneity in therapeutic outcomes, with a significant portion of patients failing to achieve functional recovery or suffering complications such as hemorrhagic transformation. Current preoperative predictive markers are insufficient; existing indicators (e.g., recanalization grade, postoperative hemorrhage) are retrospective or post-procedural, creating a time lag that prevents precise preoperative screening. The study addresses the need for novel preoperative risk markers to guide individualized treatment strategies and predict which LVO-AIS patients will derive clinical benefit from MT.

Methodology
The study employed a prospective clinical cohort design involving 22 patients with anterior circulation AIS admitted to The First Affiliated Hospital of Hebei North University between May 2025 and May 2026.

• Patient Stratification: Patients were stratified into "favorable" and "unfavorable" prognosis groups based on two endpoints: early neurological improvement (ENI ≥ 30% reduction in NIHSS at 24 hours) and long-term functional prognosis (90-day modified Rankin Scale [mRS] ≤ 2).
• Sample Collection: Peripheral venous blood was collected into PAXgene tubes prior to the endovascular thrombectomy (EVT) procedure.
• Transcriptomic Analysis: RNA sequencing (RNA-seq) was performed using the Illumina NovaSeq platform. Data processing included quality control (fastp), alignment to the human reference genome (GRCh38.p14) via HISAT2, and quantification using RSEM.
• Bioinformatics: Differentially expressed genes (DEGs) were identified using DESeq2 (criteria: |log2FC| ≥ 1, FDR < 0.05). Functional enrichment was conducted via Gene Ontology (GO) and KEGG pathway analysis. Protein-protein interaction (PPI) networks were constructed to identify hub genes.
• Statistical Modeling: Univariate and multivariate Cox regression analyses were used to identify genes associated with procedural benefit. Receiver Operating Characteristic (ROC) curves and Decision Curve Analysis (DCA) were utilized to evaluate the predictive performance and clinical net benefit of the identified gene signatures.

Key Contributions and Results

• Transcriptomic Profiling: The analysis identified 567 significant DEGs between responders and non-responders. Specifically, 432 genes were upregulated in the favorable group (enriched in immune-inflammatory pathways), while 135 were downregulated (enriched in neuronal apoptosis and endothelial injury pathways).
• Pathway Enrichment: Gene Set Enrichment Analysis revealed that the favorable group exhibited significant enrichment in inflammatory biological processes, with "cytokine-cytokine receptor interaction" identified as the top KEGG pathway. This suggests that preoperative inflammation may facilitate neurological recovery in this context.
• Hub Gene Identification: Through PPI network analysis and regression modeling, four neutrophil-related hub genes were identified as core regulators: AZU1, ELANE, MPO, and MMP8.
  • Correlation: These four genes showed strong correlation with MT efficacy. Notably, while DEFA1B showed distinct expression differences, it did not serve as an independent prognostic regulator in multivariate analysis.
  • Predictive Performance: The logistic regression model based on ELANE achieved an Area Under the Curve (AUC) of 0.893. The four-gene panel demonstrated good discriminative ability for both short-term (ENI) and long-term (90-day mRS) outcomes, with AUCs ranging from 0.785 (MMP8) to 0.832 (MPO).
  • Clinical Utility: Decision Curve Analysis (DCA) indicated that the four-gene model provided a maximum clinical net benefit of 0.64, outperforming reference strategies across relevant threshold probabilities.

Significance and Claims
The paper claims that preoperative peripheral blood transcriptomic profiles differ significantly between MT responders and non-responders, specifically regarding inflammatory gene expression. The study posits that individuals with elevated expression of specific inflammatory genes (AZU1, ELANE, MPO, MMP8) are more likely to derive clinical benefits from surgery.

The authors assert that these four hub genes serve as preliminary, non-invasive candidate predictive biomarkers. Their significance lies in:

1. Providing a molecular basis for the variable post-intervention treatment responses observed in AIS patients.
2. Offering a potential tool for preoperative risk stratification and the formulation of individualized MT strategies.
3. Suggesting that the acute systemic inflammatory response triggered by large-vessel occlusion plays a regulatory role in both short- and long-term neural repair processes.

Limitations and Modesty of Claims
The authors explicitly acknowledge the exploratory nature of the study and several limitations:

• Sample Size: The cohort was small (n=22) and single-center, limiting generalizability and the ability to adjust for multiple confounding variables.
• Causality: The study identifies correlative associations but cannot establish causal relationships between gene expression levels and surgical benefit.
• Cellular Resolution: Bulk RNA-seq was used, preventing the precise identification of specific cell subsets driving the expression; the authors note that single-cell sequencing is required to clarify cell-specific patterns.
• Validation: The findings require validation in larger, multi-center cohorts with independent external validation before clinical translation.

Consequently, the paper frames these genes as "preliminary" and "candidate" markers, emphasizing the need for further cellular, animal, and prospective clinical research to validate the regulatory mechanisms and clinical utility.