The incremental prognostic value of sIL-2R and HGF in acute ischemic stroke

Background: Inflammation affect long-term neurological outcome after ischemic (AIS). outcome may offer new biomarkers or therapeutic approaches for AIS. Methods: We collected plasma from in 204 AIS patients and 76 healthy controls, and ten cytokines (HGF, IL-1β, IL-2, sIL-2R, IL-5, IL-10, IL-16, MIP-3α, CD40L and MMP1) screened out by Immune Monitoring 65-Plex Human ProcartaPlex Panel were measured. Functional outcome 3 months after stroke was assessed using the modified Rankin Scale. To assess the prognostic ability of inflammatory mediators, we applied multivariate logistic regression and construction of multimarker score. Results: HGF, IL-10, IL-1β, MIP-3α, IL-2, sIL-2R, and IL-5 were significantly upregulated in AIS patients compared to control. After multivariable adjustment, sIL-2R (OR, 1.138; 95% CI, 1.028-1.259; P =0.012) and HGF (OR, 1.121; 95% CI, 1.030-1.218; P =0.008) remained individually associated with unfavorable outcomes at 3 months ( p < 0.05). Furthermore, adding sIL-2R and HGF to the conventional model significantly improved risk reclassification for unfavorable outcomes (continuous net reclassification improvement 32.18%, p < 0.001; integrated discrimination improvement 10.21%, p < 0.001). Conclusions: Higher plasma sIL-2R was a new independent predictor of unfavorable in AIS, and incorporation of sIL-2R and HGF into the conventional model significantly improved risk stratification for unfavorable outcomes. risk stratification pre-existing risk AIS for unfavorable outcomes. And adding HGF and sIL-2R simultaneously to the basic model offered the greatest incremental predictive capacity for the primary outcome. These findings indicated that increased plasma HGF and sIL-2R were associated with unfavorable prognosis of AIS and might be potential prognostic biomarkers for AIS. levels, triglyceride levels, total cholesterol levels, high-density lipoprotein levels and low-density lipoprotein levels.

prevention seems necessary.
Convincing evidence has implicated global brain inflammation as possibly critical factors affecting the evolution of pathology after a stroke and shaping stroke patients' long-term neurological outcomes [3][4][5][6]. Exploring the inflammatory mechanisms of ischemic stroke will undoubtedly promote rehabilitation, as illustrated by the observation that pretreatment with angiotensin-converting enzyme inhibitors is a predictor of good outcomes, owing to the proinflammatory effects of angiotensin II [7]. A number of studies have investigated relationships between inflammation biomarkers, such as Creactive protein, IL-6, TNF receptor and IL-10, and stroke outcome, and they were reported to be associated with adverse or good outcomes in AIS patients [8][9][10]. Additionally, our team reported a Th1 to Th2 lymphocyte shift in the acute phase of AIS patients and conspicuous Th1 to Th2-related cytokine alterations in the plasma of middle cerebral artery occlusion mice [11]. It was recently reported that the specific ex vivo released cytokine profile is associated with ischemic stroke outcome and improves its prediction, indicating the utility of ex vivo synthesized cytokines for predicting stroke outcome [12]. Except for cytokines, lymphocytes produce increased amounts of acetylcholine in AIS patients, which might contribute to fatal post-stroke infection and mortality [13]. It was also suggested that reduced serum levels of irisin were powerful biological markers of risk of developing post-stroke depression [14], and serum dickkopf-3 is associated with death and vascular events after ischemic stroke [15]. However, to discover the role of new inflammatory factors in the prognosis of ischemic stroke, especially the combined prognostic significance of multiple inflammatory biomarkers on stroke outcome are needed [16].
We studied a broad panel of 65 circulating inflammatory biomarkers, including T and B lymphocyterelated cytokines, immune cell infiltration-related chemokines, blood brain barrier breakdown-related factors, apoptosis-and neuronal survival-related cytokines, and angiogenesis-and neurogenesisrelated factors in AIS patients to enhance our understanding of the relationship between the involved inflammatory cytokines and adverse outcomes at 3 months after stroke events. We hypothesized that the simultaneous employment of multiple inflammatory cytokines may improve the prediction of unfavorable outcomes and improve risk stratification beyond pre-existing risk factors in AIS patients.

Methods
Invitrogen, PPX-10). Experienced laboratory technicians who performed the assays were blinded to the experimental groups, baseline characteristics and clinical outcomes of the study patients.

Outcome assessment
Outcome 90 days after stroke was represented as the mRS score. Primary outcome was an unfavorable outcome at 90 days after stroke, and mRS score 2 points indicated an unfavorable outcome. The follow-up was evaluated by trained neurologists unaware of treatment assignment.

Statistical analysis
The results of continuous data are expressed as the means ± SDs or medians (quartiles), and categorical data are expressed as percentages. Comparisons between continuous variables were carried out using Student's t test if the variables were normally distributed, or else using the Mann-Whitney U test. Comparisons between categorical variables were implemented using the χ 2 test. First, the Mann-Whitney U test was employed to identify the aberrantly expressed cytokines in AIS patients.
Then, in a verified patient cohort, the correlations between those cytokines and the study outcomes were examined using univariate and multivariable regression analyses. Crude and adjusted odds ratios (ORs) of each biomarker, along with the corresponding 95% confidence intervals (CIs) were reported. There were several potential variables in the multivariable analysis, including age, admission NIHSS score, history of diabetes mellitus, coronary heart disease and atrial fibrillation, recombinant tissue-plasminogen activator (rt-PA) treatment, leukocyte count, glucose levels, total cholesterol, low-density lipoprotein, HGF, IL-1β, IL-16, IL-2, sIL-2R and IL-5 on admission. Furthermore, receiver operating characteristic (ROC) curves were applied to calculate optimal cutoff points of HGF and sIL-2R in predicting unfavorable outcomes in AIS patients. Next, we calculated the net reclassification index (NRI) and integrated discrimination improvement (IDI) [18] to evaluate the reclassification value through adding one or more of these inflammatory biomarkers to the conventional model with existing risk factors.
All statistical analysis was conducted using SPSS 21.0 software (IBM Corp., Armonk, NY, USA) and R software (version 3.5.1). A p value 0.05 was considered significant.

Comparisons of baseline characteristics and cytokine levels between the group with good and poor outcome
The patient's characteristics and the ten aberrantly expressed cytokines, stratified by functional outcome according to the mRS score at 3 months (Table 2). Patients with unfavorable outcomes tended to be older, had higher baseline NIHSS scores, less likelihood of receiving rt-PA treatment, higher odds of atrial fibrillation history, higher baseline white blood cell counts and baseline glucose levels, and lower triglyceride levels and total cholesterol levels. Notably, we observed significant increases in HGF (138.64 versus 89.09, p < 0.000), IL-16 (87.03 versus 56.09, p < 0.000) and sIL-2R levels (1346.94 versus 812.92, p < 0.000), and significant reductions in IL-1β (4.83 versus 6.34, p = 0.024) and IL-2 levels (23.15 versus 29.74, p = 0.013) in patients with unfavorable outcomes compared with patients with favorable outcomes (Table 2, Figure 2).

Higher plasma sIL-2R and HGF are independent predictors of unfavorable outcomes in AIS
In the univariate analysis, HGF, sIL-2R, IL-16, IL-2, and IL-1β were all associated with unfavorable outcomes at 3 months after AIS (p < 0.05 for all, Table 3). After adjusting for gender, age, baseline NIHSS score and other clinical variables possibly associated with unfavorable outcomes in binominal multivariate analysis (model 2), only plasma HGF and sIL-2R remained significant for prediction of an unfavorable outcome 3 months after AIS (p < 0.05 for both). The multivariable adjusted OR (95% CIs) of each 10 pg/ml higher of HGF was 1.121 (1.030-1.218), and the multivariable adjusted OR (95% CIs) of each 100 pg/ml higher of IL-2R was 1.138 (1.028-1.259) ( Table 3).
Next, we conducted ROC curve analysis to obtain the optimal cut-off points for HGF and sIL-2R. The optimal cut-off value of HGF was 117.915 pg/ml, with a sensitivity of 65.8% and a specificity of 83.2%, and the AUC was 0.786 (95% CI 0.719-0.854). The optimal cut-off value of sIL-2R was 971.44 pg/ml, with a sensitivity of 75.3% and a specificity of 67.2%, and the AUC was 0.768 (95% CI 0.702-0.835).
We also found that HGF levels ≥ 117.915 pg/ml and sIL-2R ≥ 971.44 pg/ml were both associated with an increased risk of the primary outcome 90 days after AIS (Table 3).

Incremental predictive value of plasma sIL-2R and HGF for the prognosis of AIS
To construct multimarker score, we used two cytokines that remained significantly associated with poor outcome in the final backward elimination model: HGF and sIL-2R. We examined whether adding plasma HGF or sIL-2R to a conventional model could improve the predictive value for the prognosis of AIS. Individually, adding each of them to the conventional model consisting of risk factors in model 2 significantly improved the risk reclassification for unfavorable outcomes (p < 0.05 for both NRI and IDI) (Table 4). Furthermore, adding plasma HGF and sIL-2R at the same time to the conventional model offered the greatest incremental predictive capacity for the primary outcome (continuous NRI 32.18%, p < 0.001; IDI 10.21%, p < 0.001).

Discussion
In the present study, we first investigated 65 circulating inflammatory biomarkers simultaneously in AIS patients and their relationship with long-term neurological outcomes. We found that plasma HGF and sIL-2R were each independently associated with increased risk of unfavorable outcomes at 3 months after AIS in our study. Furthermore, adding plasma HGF or sIL-2R to traditional risk factors could improve the risk stratification for unfavorable outcomes. And adding HGF and sIL-2R simultaneously to the basic model offered the greatest incremental predictive capacity for the primary outcome. These findings indicated that increased plasma HGF and sIL-2R were associated with unfavorable prognosis of AIS and might be potential prognostic biomarkers for AIS.
Inflammation is involved in the pathological process of ischemic stroke. Few studies have globally investigated the alterations of inflammatory cytokines in AIS patients and healthy controls. We first screened 65 inflammatory cytokines in our patient cohort by Immune Monitoring 65-Plex Human ProcartaPlex Panel. Ten cytokines (HGF, IL-1β, IL-2, sIL-2R, IL-5, IL-10, IL-16, MIP-3α, CD40L and MMP1) were screened to change significantly (Supplementary Figure 1). Then we collected plasma from in 204 acute ischemic stroke (AIS) patients and 76 healthy controls, and found that seven cytokines including HGF, IL-1β, IL-2, sIL-2R, IL-5, IL-10, and MIP-3α indeed upregulated and had diagnostic capacity in AIS patients. Therefore, our first contribution was to identify two new biomarkers sIL-2R and MIP-3α, which can provide references to other studies and help target particular pathways to prevent the progression of ischemic stroke.
Then, in the univariate analyses, higher plasma HGF, sIL-2R and IL-16 levels were found to be associated with increased risk of unfavorable outcomes, while higher IL-2, and IL-1β levels were associated with decreased risk of unfavorable outcomes. Next, after adjusting for potential risk factors in binominal multivariate analysis, HGF and sIL-2R remained associated with increased risk of unfavorable outcomes, suggesting that plasma HGF and sIL-2R at baseline may be potential predictive biomarkers for prognosis of AIS. Furthermore, the addition of plasma HGF, sIL-2R or both to conventional risk factors was shown to improve risk predictions for the primary outcome. Therefore, this is the first research showed that plasma sIL-2R might be useful in risk stratification in AIS prognosis and could be beneficial for the selection of high-risk patients who should receive aggressive monitoring and therapeutic interventions in future clinical practice.
In recent years, serum HGF has emerged as a novel biomarker for cardiovascular and cerebrovascular diseases. Bielinski et al [19] reported that HGF is a biomarker of atherosclerotic disease and is associated with subclinical and incident coronary heart disease. Moreover, Susen et al [20] found that high serum HGF is an independent predictor of a composite of death and myocardial infarction. In terms of stroke, HGF was proved to be positively associated with the incidence of stroke [21,22], and Zhu et al [23] reported that higher HGF was associated with mortality but not disability at 3 months after ischemic stroke onset. However, Zhu et al included patients within 48 hours of symptom onset and excluded patients treated with rt-PA. By comparison, patients' blood in our study was collected within 24 hours before they received any treatment. Thus, the level of HGF in our study could reflect more about the authentic pathological change of AIS without the influence of other factors. Next, we included patients who received rt-PA therapy. rt-PA therapy is a widely accepted treatment strategy for AIS patients within 4.5 hours after symptom onset and can reduce patients' disability to a large extent [24]. Hence, most patients will accept rt-PA treatment if their onset time is less than 4.5 hours, and including patients with rt-PA therapy could better fit real clinical settings. Moreover, Zhu's study only ruled out AIS patients in deep coma, without considering patients whose premorbid mRS ≥ 2.
However, for patients with a severe disability, they had a higher probability of having a poor prognosis, which may result in statistical bias. Therefore, we only included patients whose premorbid mRS ≤ 1 minimize this bias. The mechanism of HGF was clarified in the basic research. Moreover, we first reported that sIL-2R was independently associated with increased risk of unfavorable outcomes in AIS. It can be seen in our data that sIL-2R is a high abundance protein in plasma. Therefore, it is easy to detect and suitable as a molecular marker. SIL-2R, a membrane receptor for IL-2, is expressed on the surface of activated T-cells and is shed into the circulation in a soluble form as sIL-2R. Previous research indicated that elevated serum levels of sIL-2R were associated with a poor prognosis in autoimmune diseases, such as multiple sclerosis and follicular lymphoma [28]. Peter et al reported that sIL-2R was positively associated with internal carotid wall thickness, cardiovascular disease mortality, incident cardiovascular disease and stroke [29]. In addition, sIL-2R was significantly higher in ischemic left ventricular dysfunction patients [30] and was associated with a worse prognosis for dilated cardiomyopathy patients [31]. Similarly, we first found that sIL-2R was positively associated with poor functional outcomes in AIS patients. We speculate that abnormal expression of sIL-2R is associated with the aberrant activation of T cells and promotion of neuroinflammation after ischemic stroke [32]. Importantly, according to our supplementary data (Tables 1 and 2), the number of neutrophils in the HGF high expression group was higher than that in the low HGF expression group, while the number of lymphocytes in the high sIL-2R expression group was lower than that in the low sIL-2R expression group. And previous studies suggested that high levels of neutrophils and low levels of lymphocytes were both associated with poor functional outcomes after AIS [33]. Above all, it is of interest to further elucidate the precise mechanisms between increased sIL-2R levels and unfavorable prognosis of AIS.
However, there are some limitations in our study. First, our study lacked data on infarct volume in MRI or CT scan. Patients who met the criterion of intravenous therapy should receive CT scans to exclude cerebral hemorrhage and should be infused with thrombolysis drugs as early as possible [34]. So nearly half of the patients in our study did not have a premorbid MRI scan. Besides, infarcts are not obvious and stable on early CT scans, and the severity of neurological dysfunction is not always proportional to the size of infarct volume; hence, it was reasonable that we did not include infarct volume in our study. Second, our study was performed mainly in Chinese individuals, and the patient sample was somewhat small, limiting the generalizability of the results to other ethnicities. Further studies with larger sample sizes are needed to verify our findings.

Conclusions
In conclusion, our study is the first to comprehensively study inflammatory cytokines in acute ischemic stroke and found that higher plasma sIL-2R was a new independent predictor of unfavorable outcomes in AIS. Additionally, adding plasma sIL-2R and HGF to conventional risk factors significantly improved risk stratification for poor outcomes in AIS patients, indicating that plasma sIL-2R and HGF may be potential prognostic markers for AIS.

Authors' contributions
HPZ and FFL prepared the study protocol; collected, analyzed, and interpreted the data; and prepared the manuscript. YYH, SJZ, LZL, ZHY, RLW, ZT, ZPH and JFF, YMZ collected the data. HPZ and FFL performed the cytometric assay and analyzed the data. QFM and YML prepared the study protocol; analyzed and interpreted the data; supervised the study. All authors read and approved the final manuscript.

Availability of data and materials
The datasets used during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
Written informed consent was obtained from each patient included in the study. The study protocol was approved by the Xuanwu Hospital of Capital Medical University.