Dear Sir:

Despite the application of endovascular thrombectomy, approximately 30% of patients with stroke encounter neurological deterioration following thrombectomy, and a significant proportion of patients face vascular events, including stroke recurrence, even with optimal antithrombotic management [1]. Recent strides in advanced histological analysis, incorporating transcriptomic analysis and immunohistochemistry, offer a promising avenue for obtaining new insights [2]. Consequently, our efforts have been directed towards identifying a set of microRNAs with prognostic significance from the thrombi of patients with stroke and evaluating their therapeutic potential through in vitro disease models.

Patients with ischemic stroke who underwent intra-arterial thrombectomy at a university hospital between February 2016 and March 2020 were included in the first cohort. MicroRNA expression was compared between patients with and without vascular events. The microRNA profiles from thrombus analysis were validated using blood samples from another cohort of patients with stroke who underwent 18F-fluorodeoxyglucose positron emission tomography to predict stroke recurrence. The therapeutic effects of the identified microRNAs were examined using an endothelial ischemic injury model. More detailed study methods are provided in Supplementary Methods.

A total of 79 thrombi from 103 patients within the thrombus analysis cohort were included (mean age=70.7±11.9 years, 33 female patients) and 21 vascular events (26.6%) were identified (Supplementary Figure 1). Microarray analysis revealed significant upregulation of 19 microRNAs, whereas 11 were downregulated in patients experiencing vascular events (Figure 1A). Principal component analysis of the microarray data revealed distinct but intermixed expression signatures between the two groups (Figure 1B). Subsequently, six microRNAs were selected for validation based on microarray analysis results and a literature review (Supplementary Table 1). Quantitative real-time polymerase chain reaction from thrombus analysis revealed significantly elevated expression of miR-93-5p among patients with vascular events (3.82±5.42 vs. 11.2±14.5, P<0.01), whereas miR-20b-5p demonstrated a tendency towards elevation (11.7±13.7 vs. 29.1±29.3, P=0.07) (Figure 1C). A detailed comparison of the clinical information and microRNA expression levels between patients with and without vascular events is presented in Supplementary Table 2. The Cox proportional hazards model did not reveal an independent relationship between miR-93-5p expression and vascular events (Supplementary Table 3).

Receiver operating characteristic curve analysis that predicted vascular events using miR-93-5p yielded insignificant results (area under the curve=0.608, 95% confidence interval=0.416-0.799; P=0.25) (Supplementary Figure 2A). After dichotomizing patients based on miR-93-5p expression levels from thrombus, Kaplan-Meier survival curve analysis demonstrated significantly shorter event-free survival time among patients with higher miR-93-5p expression (log-rank test, P=0.03) (Supplementary Figure 2B). A waterfall plot based on miR-93-5p expression highlighted the number of patients with vascular events (red arrow) or death (orange arrow) on the side with higher miR-93-5p expression (Supplementary Figure 2C). Although patients with vascular events tended to exhibit elevated miR-93-5p expression in venous blood samples obtained at the end of thrombectomy (P=0.058) (Supplementary Figure 2D), the expression levels of miR-93-5p in the blood and thrombus were not correlated (r=0.0275, P=0.90) (Supplementary Figure 2E).

Correlation analysis between microRNAs and laboratory variables indicated that miR-93-5p expression levels were positively correlated with the National Institutes of Health Stroke Scale score at hospital discharge (r=0.331, P=0.02) (Supplementary Figure 3A), e/e’ from echocardiography (r=0.321, P=0.02) (Supplementary Figure 3B), and serum parathyroid hormone levels (r=0.451, P<0.01) (Supplementary Figure 3C). Whereas, miR-93-5p expression levels were negatively correlated with serum vitamin D levels (r=-0.375, P<0.01) (Supplementary Figure 3D). Immunohistochemistry for miR-93-5p in the thrombus revealed abundant positive signals throughout the thrombus sections (Supplementary Figure 3E and F). Collectively, these data suggest that miR-93-5p is closely associated with thrombus formation.

Following the outcomes from the initial cohort, in the second cohort, we validated the expression patterns of miR-93-5p and miR-17-5p; the latter shares the same seed sequence as miR-93-5p and is a crucial member of the miR-17-92 cluster. Patients experiencing recurrent stroke exhibited elevated levels of both miR-93-5p (2.33±1.42 vs. 3.60±1.72, P=0.04) (Figure 2A) and miR-17-5p (2.21±1.19 vs. 3.24±1.68, P<0.05) (Figure 2D) than those with a stable prognosis (Supplementary Table 4). Interestingly, circulating miR-93-5p levels were positively associated with glucose uptake in the amygdala (r=0.326, P=0.01) (Figure 2B and C), whereas miR-17-5p levels were correlated with glucose uptake in the carotid artery (r=0.270, P<0.05) (Figure 2E and F).

Endothelial injury following oxygen glucose deprivation (OGD) correlated with augmented expression of miR-17-5p (P=0.03) (Figure 3A) and miR-93-5p (P=0.03) (Figure 3B) in human umbilical vein endothelial cells and human brain microvascular endothelial cells. The application of mimic or antagonistic sequences effectively altered the expression of the target microRNAs (Supplementary Figure 4). Notably, the application of their antagonistic sequences resulted in a pronounced enhancement in cell survival (OGD vs. miR-17-5p antagonist, P<0.01; OGD vs. miR-93-5p antagonist, P=0.10) (Figure 3C) along with a concomitant reduction in cytotoxic lactate dehydrogenase (OGD vs. miR-17-5p antagonist, P<0.01; OGD vs. miR-93-5p antagonist, P<0.01) (Figure 3D). Western blot analysis revealed that the antagonistic sequence of miR-17-5p significantly increased the expression of programmed death ligand 1 (PD-L1) (OGD vs. miR-17-5p antagonist, P=0.03) (Figure 3E). Additionally, the expression of phosphorylated serine/threonine protein kinases was significantly increased by the antagonistic sequences of both miR-17-5p and miR-93-5p (OGD vs. miR-17-5p antagonist, P=0.03; OGD vs. miR-93-5p antagonist, P=0.03) (Figure 3E).

This study demonstrates that the expression signature of microRNAs from thrombi may be a potential predictor of future vascular events after stroke. The elevation of miR-93-5p was found to be associated with future vascular events among survivors of stroke by investigating two different specimens, thrombus and blood, from two distinctive prospective cohorts. Using in vitro models, the antagonistic sequences of miR-93-5p and miR-17-5p were found to mitigate endothelial cell death, highlighting their potential roles in the intricate processes of thrombus formation.

The impact of miR-93-5p on future vascular events appears to be multifaceted given its positive association with brain activity in the amygdala, a hub for emotional, hormonal, and circadian homeostasis control. Recent studies revealed that decreased PD-L1 in thrombi or atheromas is associated with future vascular events [3,4]. This study underscores that elevated amygdala activity correlates with increased systemic expression of the miR-17/92 family, potentially attenuating a key modulator of the adaptive immune system and promoting thromboembolic tendencies among survivors of stroke [5]. However, the results of this study should be interpreted with caution because the relationship between microRNAs and vascular events was not independent after adjusting for clinical variables in the first cohort, and the correlation between miR-93-5p and amygdala activity was weak in the second cohort. This suggests that miR-93-5p may act as a confounding factor indirectly related to vascular events.

This study has several limitations. First, the sample size was relatively small and microRNA analysis proved unfeasible in approximately 20% of the harvested thrombi. Second, while the modulatory effects of the identified microRNAs were explored using in vitro models, the absence of in vivo investigations hindered a comprehensive understanding of the pathophysiological role and systemic impact of miR-93-5p in the link between thrombosis and immune reactions. Whether a therapeutic strategy to mitigate microRNA expression or its downstream pathways can prevent future vascular events warrants further study.