Characteristics of the study

Studies were published between 2017–2021. The studies were conducted in China (42.85%), USA (28.57%), Japan (9.52%), Poland (4.76%), Turkey (4.76%), and Spain (9.52%), and the sample size varied from 6 patients to 2,763 patients. All studies enrolled patients with stroke and 11 studies (52.38%) were matched with a control cohort of healthy individuals.

The combined mean and standard deviation (SD) for the age of the stroke patients was 63.86 (SD = 13.98) and in the control group was 62.047 (SD = 12.21). Two of the studies (9.52%) did not provide the age of the patients enrolled and two (9.52%) provided only an age interval (49–80 and 64–85, respectively), for which statistical analysis could not be performed.

For the patients with stroke, the male cohort was larger than the females (55.43% vs. 44.56%). This was maintained for the studies that had a control group (57.29% vs. 42.70%). Two of the studies (9.52%) included samples from databases, for which characteristics of the patients were not provided.

In nine studies, the blood samples were collected during the acute phase (< 24 h from the onset) (42.85%). Two studies collected the blood samples in the first two weeks (9.52%), two in a time frame of 120 days (9.52%), one collected the blood samples at 6 months (4.76%), and seven did not provide a specific window (33.33%). Two studies were indexed in journals with an IF under 1 (9.52%), two were indexed in journals with an IF between 1 and 3 (9.52%), eleven were in journals with an IF between 3 and 5 (52.38%) and six were indexed in journals with an IF over 5 (28.57%). Over 100 biomarkers were identified in all the studies. The characteristics of the studies can be found in Table 1.

Methodological assessment

The REMARK criteria [15] are provided in Table S2, S3, and Figure 2. Overall, two studies (9.52%) met all the REMARK criteria, two (9.52%) met 87.5%, ten (47.61%) met 75%, five (23.80%) met 62.5%, and two (9.52%) were under 50%. All studies provided a precisely defined clinical outcome and the methods of measuring the biomarker, nineteen studies (90.47%) were prospective, eighteen (85.71%) provided characteristics of the patients, fifteen (71.42%) mentioned a defined time period for enrolment, five (23.80%) uses a blinded biomarker measurement, and only two studies (9.52%) provided a rational for sample size.

Display full size

Figure 2. 

Quality study assessed by REMARK criteria (Tables S2 and S3)

Biomarkers

The genetic biomarkers associated with stroke are summarized in Table 2. Five of the studies investigated the levels of circular RNA (circRNA) in association with stroke onset or recurrence (23.80%), four focused on long non-coding RNA (lncRNA) involved in immunity/inflammatory processes (19.04%), six investigated the polymorphisms associated with the atherosclerotic process and stroke recurrence (28.57%), three inquired into DNA methylation (DNAm) processes, stroke onset and outcome (14.28%), and three studies did proteomics/genomics analysis to correlate levels of plasma proteins with pathways of stroke (14.28%).

NcRNA biomarkers of IS

The role of ncRNA has been long discussed in the pathogenesis of different pathologies. Recent studies have been focusing on the function of circRNAs and their potential as biomarkers [19]. A comprehensive review of the biogenesis of circRNAs suggests that circRNAs can operate on protein scaffolds, interact with other RNA species, and modulate transcriptional silencing, translation, and degradation of certain mRNAs [19]. Thus, the circPHKA2 (hsa_circ_0090002) and circBBS2 (hsa_circ_0039457) mean values were statistically significantly increased in the IS group compared to the sham group (student’s t-test; P < 0.05) [20]. The PHKA2 gene encodes PbK-subunit alpha. PbK activates glycogen phosphorylase, which degrades glycogen when glucose is depleted, and hence plays a crucial role in cellular energy metabolism [21]. The BBS2 gene is a part of the BBS family, which is reported to stimulate ciliary biogenesis via GTPase Rab8 [22]. However, the function of BBS2 after ischemia in the brain remains unclear.

Another recent study focused on the potential of circRNAs with regard to stroke etiology. The authors reported over 200 diff in the expression of circRNAs between cardioembolic and atherothrombotic stroke and validated hsa_circRNA_102488, generated from pre-mRNA encoded by the UBA52 gene on chromosome 19. Indeed, the levels of hsa_circRNA_102488 were lower in the atherothrombotic group (P < 0.01) and the corresponding levels of UBA52 mRNA were also lower (P < 0.001) [23]. Functionally, hsa_circRNA_102488 has a binding site for six distinct miRNAs involved in fatty acids biosynthesis, lysine degradation, and ARVC. Another study focusing on circRNAs aimed to investigate the association between circRNA levels and the onset of stroke. The study identified five miRNAs with fluctuating patterns that correlated with the onset and time course of cerebral infarction: miR-505-5p, miR-1255b-5p, miR-550b-2-5p, miR-4523, and miR-6795-3p. Higher levels of miR-376a-3p and downregulation of miR-3184-5p were associated with the onset of cerebral infarction [24].

miRNAs are key post-transcriptional regulators that regulate target genes by binding to multiple mRNAs. In cases of AIS, miRNAs have also been discovered to be critically involved in leukocyte gene expression [25]. Vijayan et al. [26] reported 16 dysregulated miRNAs in the IS patients vs. controls, out of which eight were validated by RT-PCR: four upregulated [PC-3p-57664 (P = 0.01), PC-5p-12969 (P = 0.04), hsa-miR-122-5p (P = 0.01), and hsa-miR-211-5p (P = 0.03)] and four downregulated [hsa-miR-22-3p (P = 0.01), PC-3p-32463 (P = 0.0001), hsa-miR-30d-5p (P = 0.0009), and hsa-miR-23a-3p (P = 0.03)]. However, it was not clear if they act as biomarkers of stroke risk factors or are a consequence of cerebral ischemia.

Exosome-derived ncRNAs implicated in inflammation and fibrosis, notably miRNAs, have been found to modify organ-level phenotypes in animal models of stroke [27]. Because exosome-derived ncRNAs are released into circulation and may act as distant molecular mediators, there has been an increasing amount of research seeking to define a plasma exosome-derived ncRNA signature of systemic disorders. Thus, Mick et al. [28] found that exosome-derived ncRNAs were each independently associated with stroke incidence: hsa-miR-877-5p (odds ratio, 0.18; 95% CI, 0.06–0.52; P = 0.002), hsa-miR-124-3p (odds ratio, 0.29; 95% CI, 0.11–0.72; P = 0.008), hsa-miR-320d (odds ratio, 0.33; 95% CI, 0.14–0.78; P = 0.01), and snoRNA [SNO1402 (odds ratio, 0.20; 95% CI, 0.07–0.52; P = 0.001)]. In each case, individuals with a history of stroke at baseline were less likely to express the exosome-derived ncRNA at baseline than those without a history of stroke at baseline. Also, employing several Cox-proportional hazards models, they reported that hsa-miR-656-3p and hsa-miR-941 absolute expressions were independent of stroke risk and significantly associated with a higher risk of stroke [28]. Several studies found that serum/plasma lncRNA levels are associated with post-stroke immunity/inflammatory processes [29]. Thus, the involvement of lncRNA in the transition from acute to subacute stroke was investigated at 24 h and 7 days post-stroke by Gene Ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis. The findings highlighted the role of peripheral immunity in AIS, with three lncRNAs (SCARNA10, TERC, and LINC01481) showing significantly increased expression (P < 0.01), whereas FLJ23867, H3F3AP6, and TNPO1P1 showed significantly decreased expression (P < 0.05) in PBMCs from IS patients compared to healthy controls [29].

A LncMRT is a functional network motif, made up of a lncRNA, a TF, and a gene that governs various aspects of life [30]. Cao et al. [31] used a global LncMRT network to identify their ability as new biomarkers and treatment targets for IS. Dysregulated LncMRTs in IS allowed the identification of the PI3K/Akt signaling pathway as a critical pathway associated with stroke onset and recurrence. The dysregulated LncMRT in IS patients (consisting of PTEN, TFAP2A, and HOTAIR) suggests that PTEN may play a role in the PI3K/Akt signaling pathway by generating LncMRT. Indeed, serine/threonine kinase PI3K is associated with improved cell survival, and this effect is mediated in part by the activation of the serine/threonine kinase Akt [32].

Another recent study aimed to identify the role of lncRNAs and mRNAs in a cohort of IS patients and healthy controls [33]. Five differentially expressed lncRNAs (LNC_000015, LNC_001727, SNHG8, MIRLET7BHG, AF001548.5) were found differentially expressed in IS vs. controls and used to construct a GO/KEGG pathway network that suggests their potential as biomarkers and therapeutic targets in IS-linked inflammatory response.

Several in vitro and in vivo investigations have already shown that lncRNAs associated with inflammation may be potential targets for the diagnosis and treatment of AIS [34–36]. By acting as ceRNAs for miRNAs, which are ncRNAs of about 20 nucleotides in length that negatively impact the expression of target genes at the post-transcriptional level, lncRNAs may play a part in the development of diseases [37]. Thus, Zhang et al. [38] identified multiple inflammatory associated lncRNAs, including DELs, MCM3AP-AS1, LINC01089, ITPK1-AS1, and HCG27, which may act as biomarkers for the diagnosis and treatment of AIS. These DELs (MCM3AP-AS1/LINC01089/hsa-miR-125a/FYN, ITPK1-AS1/hsa-let-7i/RHOA/GRB2/STAT1, and HCG27/hsa-miR-19a/PXN) may function as a ceRNA network to accelerate the expression of the genes. Contrary to other reports [39], HCG27 was found to be upregulated in IS. This finding may be explained by different sample selection (chronic vs. acute) and patient cohorts. The correlation of these lncRNAs’ interaction with miRNA-mRNA pairs and AIS suggests their participation in AIS. However, this study did not give any direct evidence to support it [39].

Genetic polymorphisms as biomarkers of IS

This study indicates that genetic polymorphisms associated with the atherosclerotic process and stroke recurrence account for 27.27% of serum genetic biomarkers. Thus, Zhu et al. [40] in a prospective investigation with 42 months follow-up aimed to investigate polymorphisms of five immunoinflammatory genes [C-X-C motif chemokine receptor 2 (CXCR2) rs1126579, Toll-like receptor 4 (TLR4) rs11536889, ADIPOR2 rs12342, matrix metalloproteinase-2 (MMP-2) rs7201, and MMP-9 rs1056628] in association with the recurrence of IS. They report that ADIPOR2 rs12342 was significantly associated with stroke recurrence (GG vs. AA: P = 0.025; GA vs. AA: P = 0.011; GG/GA vs. AA: P = 0.004), but not with the onset age [41].

Gap junction protein is encoded by the Cx37 gene (also known as GJA4). Cx37 C1019T polymorphism causes the conversion of proline to serine, which results in the dysregulation of endothelial nitric oxide synthase (eNOS) expression and activity [42]. Since Cx37 is mostly expressed in vascular endothelial cells, monocytes, and macrophages, the most significant cells in the process of arteriosclerosis, Cx37 C1019T polymorphism influences endothelial function and elasticity eventually leading to atherosclerosis [43]. Indeed, Li et al. [44] revealed that Cx37 SNP rs1764390 and rs1764391 genotypes are associated with an increased risk of IS, and the G allele of rs1764390 was also proven to be a risk factor for IS. Another study tested the hypothesis that sex-specific genes in peripheral blood are associated with the progression of IS and can serve as potential biomarkers to predict outcome and sex-specific peripheral immunological response [45]. They showed that there are 10 upregulated genes with mostly ribosomal and protein synthesis-related functions. The downstream translation level is more significantly affected by the changes in the transcription level of ribosome-related genes. Therefore, ribosome profiling was used in an in vitro assay to investigate how oxygen and glucose deprivation (OGD) can immediately impact the transcription and translation of the genes in brain cells and they came to the conclusion that OGD had a larger effect on translation than transcription.

Thus, the prognostic ribosomal biomarkers RPS14, RPS15A, RPS24, FAU, RPL27, RPL31, RPL34, RPL35A, RSL24D1, and EEF1B2 could become novel therapeutic targets for IS. Of note, women are more likely to experience stroke-induced immunodeficiency syndrome (SIDS), and the potential mechanism could be the sex diff in the inflammatory response of ribosomes in stroke-induced peripheral immunosuppression [45].

HDAC9 found on chromosome 7p21, is a class IIa HDAC that affects the chromatin state by removing acetyl groups [46]. Shroff et al. [47] hypothesized that vascular atherosclerosis, which is an important cause of IS, is associated with HDAC9 polymorphism rs2107595. However, neither rs2107595 risk allele-positive and risk allele-negative variants, as well as the risk allele-positive and risk allele-negative VRFC groups, did not significantly differ from each other.

Recurrent pathways involving interleukin-6 (IL-6) signaling, extracellular signal-regulated kinase (ERK)/mitogen-activated protein kinase (MAPK) signaling, liver X receptors (LXR)/retinoid X receptor (RXR) activation, and the function of macrophages, fibroblasts, and endothelial cells in rheumatoid arthritis were identified using pathway analysis for the 155 genes associated with rs2107595 polymorphism. Indeed, leukocyte recruitment, chronic inflammation, immunological response of macrophages, cholesterol efflux, and platelet aggregation were all connected to SNP rs2107595 and the risk of LVAS. However, this risk allele did not change significantly the levels of HDAC9 expression. LVAS is not induced by the risk allele rs210759 either. Thus, many individuals who carry the risk allele do not suffer a stroke. Nevertheless, the HDAC9 polymorphism may interact with other factors to increase the risk of stroke. A polymorphic allele at some other locus, epigenetic regulation, or external factors could all have an impact on the HDAC9 polymorphism’s incomplete penetrance [47].

Williams et al. [48] analyzed the involvement of plasma vWF in the recurrence of IS using expression quantitative trait locus (eQTL) analysis. Their results suggest that high vWF levels associate with the risk of stroke recurrence in both unadjusted (P = 0.0007; HR = 1.26) and fully adjusted (P = 0.018; HR = 1.19) models, with the most strongly associated SNP rs505922 [48].

A metabolomic-epigenetics study investigated the stroke recurrence in an African American population [49]. Thus, the alkaline ceramidase 2 (ACER2) locus encoding ACER2, sphingosine, and sphinganine were associated with five common variants and one different variant [rs10757056 (sphingosine only), rs10118089, rs7020745, rs10125228, rs10757058, and rs10118371]. The most significant association included leukotriene B4 and rs7025659-SPTLC1 (P = 6.12e-07; post-hoc comparison between TG to GG genotypes, P = 3.00E-07) [49]. Further, individuals with IS and high leukotriene B4 levels were linked to less successful functional recovery [50].

In a follow-up study, Davis Armstrong et al. [51] investigated DNAm loci associated with stroke recurrence in a cohort of African descendants and European descendants. They found that cg03584380 [HR = 5.41 (2.91–10.06)] polymorphism was significantly associated with the recurrence of stroke in the African population, but not with the European cohort. This methylation site is located in the first intron of the ZDHHC6 gene, which encodes palmitoyltransferase ZDHHC6 and is necessary for the palmitoylation of several important ER proteins, such as calnexin and the ITPR1 [52].

DNAm as a biomarker of IS

DNAm is an epigenetic pathway that influences the expression of genes and greater-order DNA structure. At a cytosine-phosphate-guanine (CpG) locus, the heritable but also reversible addition of a methyl group to the 5-carbon position of cytosine is linked with gene suppression [53]. DNAm levels fluctuate over the length of a human’s life and are affected by both genetic and lifestyle variations [54–56]. Age-related changes in DNAm are commonly accepted and two recent studies used methylation measurements at several CpGs loci across the genome to determine chronological age (C-age) in humans [57, 58]. Thus, Soriano-Tárraga et al. [59] found that an epigenetic biomarker calculated from DNAm values outperformed chronological aging in predicting mortality three months after an IS. This connection is particularly important in the etiology of LAA stroke [59]. Their findings are consistent with earlier research on many disorders, which has demonstrated that this biomarker accurately predicts death regardless of lifestyle, genetic factors, or health condition [60, 61].

A number of neurodegenerative disorders are significantly influenced by DNAm [41] and RIN3 gene is overexpressed and hypomethylated in the blood of patients with Alzheimer’s disease [62, 63]. Miao et al. [64] looked for the dysregulation of RIN3 gene methylation in association with cognitive impairment secondary to transitory or mild cerebral ischemia. In the TIA/MIS group, the RIN3 gene’s overall methylation level was considerably lower than in the control group (diff = 0.02; P = 0.001), the result that was maintained in the cognitive impairment vs. non-cognitive impairment group (diff = −0.013; P = 0.01), concluding that methylation levels of RIN3 could be used as a predictor factor for early cognitive impairment in IS.

Transcriptomic and proteomic biomarkers of IS

Another proteomic study aimed to identify diff between stroke subtypes and cystathionine β-synthase (CBS) deficiency those innated deficit is associated with thromboembolism [65]. Regarding stroke subtypes, nine proteins uniquely differentiated cardioembolic and large-vessel stroke (APCS, APOM, C1qA, C4BPA, C4BP2, FBLN1, IGKV1D-12, KLKB1, SERPINF2). Another six were specific for large-vessel and lacunar stroke (APOL1, C5, GPX3, GSN, H2AFJ, IGK), and six differentiated cardioembolic and large-vessel stroke (AMBP, APOA4, FCN3, ITIH4, LBP, PF4). Furthermore, Ingenuity Pathway Analysis software (IPA; Ingenuity Systems; Mountain View, CA, USA) was used to identify proteins and biological pathways linked to the clothing cascade (A2M, FBLN1, FGA, PLG, SERPIND1, F13B, etc.), the acute phase response (SAA1, CRP, SERPINA3, SERPINA1, LBP, ORM1, AHSG), and lipid metabolism (APOC1, APOC3, APOL1, etc.).

Tufekci et al. [66] aimed to verify if serum TRAIL protein levels and mRNA expression in PBMCs of stroke patients were linked with stroke prognosis throughout the course of 6 months follow-up.

It is known that TRAIL is linked to five different receptors, two of which, DR4 and DR5 causing apoptosis and inflammation [67]. On the onset, blood levels of TRIAL protein were lower in the stroke cohort (P = 0.0001), but with a higher TRAIL mRNA expression in PBMC. In the follow-up period, the authors reported a down-regulation of the protein levels, while TRAIL mRNA expression in PBMC was found to be upregulated. However, no statistical significance was seen when blood TRAIL protein levels and PBMC TRAIL mRNA expression were compared among stroke subtypes, concluding that levels of serum TRIAL protein or PBMC TRAIL mRNA expression could not be used as a potential diagnostic biomarker.

Path analysis [68] indicated a strong correlation between anti-AP3D1-antibody levels and maximum intima-media thickness, suggesting that this marker accurately captured the initiation of atherosclerosis. It was discovered that the target antigen recognized by serum IgG antibodies in the plasma of atherosclerosis patients was AP3D1. The findings demonstrated that s-AP3D1-Ab levels were substantially greater in patients with IS than in patients with healthy donors. The levels of s-AP3D1-Ab were still considerably greater in patients with AIS than in healthy donors even after the subjects’ ages were matched to 65 years. The s-AP3D1-Ab-positive rates in healthy donors and patients with AIS and TIA were 2.4%, 10.1%, and 10.4%, respectively, at a cutoff value comparable to the average plus two SDs of the HD values.

Strengths and limitations

A strength of this systematic review is the search for genetic biomarkers, which allowed us to investigate the less known potential biomarkers of stroke. Moreover, further classification of types of genetic biomarkers shall give a more clear overview of what is the main focus of recent research. On the other hand, the results could have been impacted negatively by unknown biomarker patterns and/or inconsistent blood collection times within and between experiments. Future prospective studies on the expression of biomarkers after IS should take these issues into consideration. Another drawback is the small sample sizes in some of the studies and the rationale of the sample size calculation was not taken into account, which inevitably weakens the statistical strength. Also, the study did not specifically take into account the recovery association of all the biomarkers, it was found to be associated with stroke incidence and functional recovery, which means that the data available are mixed between potential recovery prognostic genetic biomarkers and the onset of stroke.

In conclusion, more than 100 potential biomarkers were found and the data suggest that combinations of plasma genetic biomarkers might be used as a better predictor for stroke.

To the best of our knowledge, this is the first systematic review to focus only on genetic biomarkers. More than 100 biomarkers were significantly associated with IS, classified in circRNAs, lncRNA, polymorphisms, and genetic/proteomic networks. The data found suggest that genetic plasma biomarkers might be a useful tool with which to assess the risk of stroke onset and to allow prognosis of recovery in the clinic. However, more studies are needed with wider sample sizes and more homogeneous results to allow the identification of reliable plasma genetic factors unambiguously associated with stroke.