Abstract
Introduction: There are very limited data on the role of biomarkers correlating with the outcome in acute ischemic stroke (AIS). We evaluated the predictive values of the plasma concentrations of soluble serum stimulation-2 (sST2), matrix metalloproteinase-9 (MMP-9), and claudin-5 in AIS. Methods: The biomarker levels in the plasma samples of consecutive AIS patients collected at baseline, 12 h, and 24 h from stroke onset were quantified using immunoassays. Stroke severity was assessed using the National Institutes of Health Stroke Scale (NIHSS) and functional outcome at 90 days using the modified Rankin Scale (mRS), with scores above 3 defined as poor outcome. Receiver operating characteristic curve analysis and multiple logistic regression were performed for evaluating the discriminative power of each marker. Results: We included 108 patients in the study (mean age 62.3 ± 11.7 years). Median NIHSS score was 12 (interquartile range 8–18). High baseline glucose levels, systolic blood pressure, baseline NIHSS, low Alberta Stroke Program Early CT Score, and hemorrhagic transformation were associated with poor outcomes. Elevated sST2 at 12 h (50.4 ± 51.0 ng/mL; p = 0.047) and 24 h (81.8 ± 101.3 ng/mL; p = 0.001) positively correlated with poor outcomes. MMP-9 (p = 0.086) and claudin-5 (p = 0.2) were not significantly associated with the outcome, although increased expressions of both markers were observed at 12 h. Multiple logistic regression showed that sST2 levels ≥71.8 ng/mL at 24 h, with a specificity of 96.9%, emerged as an independent predictor of poor functional outcome (OR: 6.44; 95% CI: 1.40–46.3; p = 0.029). Conclusion: Evaluation of sST2 may act as a reliable biomarker of functional outcome in AIS.
Introduction
The short-term functional outcome of acute ischemic stroke (AIS) is predicted by risk factors such as age, sex, diabetes, hypertension, atrial fibrillation, hyperlipidemia, and stroke severity [1]. A reliable prognostic biomarker system adjunctive to routine assessment may aid in decision-making for the management and rehabilitation of AIS.
Previous biomarker studies have shown that circulatory proteins expressed during the postischemic mechanism of inflammation and the blood-brain barrier (BBB) disruption [2] may predict outcomes and mortality in 90 days [3‒8]. However, the clinical utility of biomarkers remains elusive due to the contradictory findings in other studies. This may partly result from the heterogeneity in biomarker assessment across studies as their expression is time-bound. Assessing at key time points from stroke onset may appropriately identify potential markers and candidate pathways.
Matrix metalloproteinase-9 (MMP-9), soluble serum stimulation-2 (sST2) protein, and tight junction proteins are associated with the disruption of the BBB and poststroke inflammation leading to worse outcomes in patients. Inflammation directly influences the sustenance of the BBB by causing a fivefold increased expression of the neutrophil-borne MMP-9 and exacerbates the tight junction protein degradation such as claudin-5 and occludin, which leaks out during barrier disruption and is expressed in the blood [9, 10].
sST2 protein is an emerging biomarker in the progression of brain injuries, highly expressed in microglia and astrocytes which are involved in neuroinflammation [11]. It is a decoy receptor for IL-33, an IL-1-like cytokine expressed in cells constituting the BBB, and endothelial cells [11, 12].
In this study, we sought to determine the discriminative values of MMP-9, sST2, and claudin-5 to prognosticate functional outcomes in a South Indian cohort of AIS patients. We hypothesized that (i) based on their role in the postischemic pathways, all three biomarkers will be elevated in the plasma at specific time points, and (ii) depending on their pathophysiological roles, each biomarker may correlate with the outcome.
Materials and Methods
Study Population
This study received ethical approval to be conducted at Sree Chitra Tirunal Institute for Medical Sciences and Technology, a tertiary care and academic medical center in Kerala, India, and the study protocol was approved by the institute’s Ethics Committee on human research. Each patient had given their written informed consent before being enrolled in the study. Inclusion criteria for the patients were as follows: (i) ≥18 years of age, (ii) first-ever ischemic stroke, (iii) admitted <12–14 h of symptom onset, and (iv) absence of hemorrhagic transformation (HT) on baseline brain CT or MRI. We excluded patients based on (i) baseline serum creatinine levels >2 mg/dL, (ii) use of anticoagulants, (iii) absence of central nervous system diseases, (iv) signs of concomitant infection and systemic inflammation, and (v) any malignant diseases. The patients were recruited during admission at the comprehensive stroke unit of our institute and we included patients who underwent intravenous thrombolysis, endovascular therapy, and bridging therapy as well as those who were not thrombolysed.
Clinical Factors and Outcome
The demographics, vascular risk factors, medical history, and diagnostic workup were documented. To further demonstrate the effect of systolic blood pressure on the outcome, we dichotomized all values of systolic blood pressure above and below 180 mm Hg. Stroke subtypes were etiologically classified using the Trial of ORG 10172 in Acute Stroke Treatment (TOAST) [13]. Stroke severity was documented using the National Institutes of Health Stroke Scale (NIHSS) with more than 16 points considered as moderate to severe impairment. Functional outcome at 90 days was assessed using the modified Rankin Scale (mRS) with patients who were categorized with scores of 3–6 as poor outcomes. CT or MRI was performed on all patients upon admission and at 24–48 h after admission. HT was documented using CT or MRI imaging and was defined according to the European Cooperative Acute Stroke Study criteria [14].
Biochemical Analysis
Plasma samples were evaluated at 3 time points: (i) at the time of admission or before revascularization in patients who underwent intervention, (ii) at 12 h, and (iii) 24 h from stroke onset. Ethylenediaminetetraacetic acid plasma was separated by centrifugation at 2,500 rpm for 15 min and stored as aliquots at −80°C until analysis. Each marker was assessed to determine its cut-off value in each group and was evaluated by commercially available enzyme-linked immunosorbent assays (ELISA). MMP-9 was assessed using Quantikine ELISA Human Immunoassay Kit (DM900; R&D Systems, Minneapolis, MN, USA). sST2 levels were determined using Duoset ELISA kits (DY523B-05; R&D Systems, Minneapolis, MN, USA) and claudin-5 was assessed using Cusabio ELISA kits (Cusabio Technology LLC, Houston, TX, USA). The absorbance of the analytes was measured at 450 nm using a microplate reader (Biotek ELX 800).
Statistical Analysis
The sample size was calculated based on the assumptions for one of the biomarkers, namely, MMP-9 as 75% and 24.6% above the cut-off value as reported in previous literature [15]. To achieve a power of 90% with an alpha error of 5%, the minimum sample size required was approximately 25 patients in each group. By adjusting for 4–5 confounding variables, another 50 samples, 25 in each group, were added to get a minimum of 100 patients. Continuous variables were expressed as means and standard deviation or medians and interquartile range (IQR) based on their distribution, and categorical variables were given in percentages. The predictive discrimination of each marker was first analyzed by univariate analysis. Pearson’s χ2 test, Wilcoxon rank sum test, and Fisher’s exact test were conducted for bivariate analyses. Pearson product-moment correlation was conducted to determine the relationship between the variables and biomarker levels associated with poor outcomes. Based on the intercorrelation of these markers and statistical significance, we created a multiple logistic regression model using covariates identified as statistically significant in bivariate analyses. Separate models were developed for the 3 time points of marker sampling. The logistic regression model including the biomarkers that provided a good discriminative capacity to predict functional outcomes was finally included. Receiver operator characteristic (ROC) curves and area under the curve (AUC), in which sensitivity was plotted as a function of (1-specificity), were used to identify optimal cut-off levels and compare predictive accuracies of the markers. A probability value less than 0.05 was considered to be statistically significant. The analysis was carried out using R version 4.2.0 software.
Results
Between December 2018 and April 2021, 111 consecutive patients were enrolled in the study. Three patients were lost to follow-up resulting in 108 patients at the 90-day follow-up. A total of 11 (9.9%) patients had died within 3 months of the stroke event. The flow diagram of the study population is provided in Figure 1. The baseline characteristics of the patient population based on their functional outcome at 90 days are given in Table 1.
Baseline characteristics of the study population
Characteristic . | Good outcome at 90 days, N = 65 . | Poor outcome at 90 days, N = 43 . | p value1 . |
---|---|---|---|
Male, n (%) | 45 (69) | 30 (70) | >0.9 |
Age, in years, mean (SD) | 62.0 (11.2) | 62.2 (12.5) | 0.9 |
Onset-to-door time, min, mean (SD) | 256 (220) | 281 (227) | 0.5 |
Hypertension, n (%) | 38 (58) | 33 (77) | 0.050 |
Diabetes mellitus, n (%) | 27 (42) | 26 (60) | 0.054 |
Coronary artery disease, n (%) | 11 (17) | 10 (23) | 0.4 |
Hypercholesterolemia, n (%) | 19 (29) | 12 (28) | 0.9 |
Atrial fibrillation, n (%) | 8 (12) | 11 (26) | 0.076 |
Rheumatic heart disease, n (%) | 7 (11) | 0 (0) | 0.040* |
Tobacco use, n (%) | 26 (40) | 13 (30) | 0.3 |
Alcohol use, n (%) | 20 (31) | 10 (23) | 0.4 |
Single antiplatelets, n (%) | 5 (7.7) | 3 (7.0) | >0.9 |
Dual antiplatelets, n (%) | 2 (3.1) | 4 (9.3) | 0.2 |
Antihypertensives | 19 (29) | 12 (28) | 0.9 |
Statins | 5 (7.7) | 3 (7.0) | >0.9 |
Plasma glucose at admission, mg/dL, mean (SD) | 158.0 (58.6) | 196.7 (91.0) | 0.016* |
Plasma glucose >140 mg/dL, n (%) | 34 (52) | 31 (74) | 0.026* |
HbA1c, n (%), median (IQR) | 6.3 (5.6, 8.0) | 7.6 (6.2, 9.8) | 0.008** |
Total cholesterol, mg/dL, mean (SD) | 193.5 (48.3) | 189.2 (64.4) | 0.2 |
HDL, mg/dL, mean (SD) | 48.6 (14.7) | 45.6 (10.7) | 0.5 |
LDL, mg/dL, mean (SD) | 127.4 (39.7) | 122.2 (56.3) | 0.2 |
Triglycerides, mg/dL, mean (SD) | 98.1 (53.2) | 98.6 (42.2) | 0.6 |
ESR, mm/h, mean (SD) | 22.7 (22.2) | 39.1 (36.0) | 0.038* |
Systolic blood pressure, mm Hg, mean (SD) | 151.3 (23.6) | 160.8 (30.6) | 0.2 |
Systolic blood pressure (≥180 mm Hg), mean (SD), n (%) | 10 (15) | 14 (33) | 0.030* |
Diastolic blood pressure, mm Hg, mean (SD) | 85.9 (13.7) | 91.1 (15.3) | 0.13 |
Baseline ASPECTS score <6, n (%) | 15 (27) | 20 (56) | 0.007** |
TOAST classification, n (%) | |||
Cardioembolism | 16 (25) | 12 (28) | 0.039* |
Large vessel disease | 17 (26) | 18 (42) | |
Small vessel disease | 8 (12) | 0 (0) | |
Stroke of other determined etiology | 1 (1.5) | 2 (4.7) | |
Stroke of undetermined etiology | 23 (35) | 11 (26) | |
IV rt-PA, n (%) | 34 (52) | 9 (21) | 0.001** |
EVT, n (%) | 17 (26) | 18 (42) | 0.088 |
Bridging Therapy, n (%) | 5 (7.7) | 1 (2.3) | 0.4 |
HT detected in patients, n (%) | 11 (17) | 17 (40) | 0.009** |
Baseline NIHSS score, median (IQR) | 10.0 (7.0, 15.0) | 17.0 (11.0, 20.0) | 0.003** |
NIHSS score at discharge, median (IQR) | 2.0 (0.0, 6.2) | 12.0 (7.5, 15.5) | <0.001*** |
mRS at discharge (3–6), n (%) | 31 (48) | 41 (95) | <0.001*** |
Mortality, n (%) | 0 (0) | 11 (25.5) | <0.001*** |
Characteristic . | Good outcome at 90 days, N = 65 . | Poor outcome at 90 days, N = 43 . | p value1 . |
---|---|---|---|
Male, n (%) | 45 (69) | 30 (70) | >0.9 |
Age, in years, mean (SD) | 62.0 (11.2) | 62.2 (12.5) | 0.9 |
Onset-to-door time, min, mean (SD) | 256 (220) | 281 (227) | 0.5 |
Hypertension, n (%) | 38 (58) | 33 (77) | 0.050 |
Diabetes mellitus, n (%) | 27 (42) | 26 (60) | 0.054 |
Coronary artery disease, n (%) | 11 (17) | 10 (23) | 0.4 |
Hypercholesterolemia, n (%) | 19 (29) | 12 (28) | 0.9 |
Atrial fibrillation, n (%) | 8 (12) | 11 (26) | 0.076 |
Rheumatic heart disease, n (%) | 7 (11) | 0 (0) | 0.040* |
Tobacco use, n (%) | 26 (40) | 13 (30) | 0.3 |
Alcohol use, n (%) | 20 (31) | 10 (23) | 0.4 |
Single antiplatelets, n (%) | 5 (7.7) | 3 (7.0) | >0.9 |
Dual antiplatelets, n (%) | 2 (3.1) | 4 (9.3) | 0.2 |
Antihypertensives | 19 (29) | 12 (28) | 0.9 |
Statins | 5 (7.7) | 3 (7.0) | >0.9 |
Plasma glucose at admission, mg/dL, mean (SD) | 158.0 (58.6) | 196.7 (91.0) | 0.016* |
Plasma glucose >140 mg/dL, n (%) | 34 (52) | 31 (74) | 0.026* |
HbA1c, n (%), median (IQR) | 6.3 (5.6, 8.0) | 7.6 (6.2, 9.8) | 0.008** |
Total cholesterol, mg/dL, mean (SD) | 193.5 (48.3) | 189.2 (64.4) | 0.2 |
HDL, mg/dL, mean (SD) | 48.6 (14.7) | 45.6 (10.7) | 0.5 |
LDL, mg/dL, mean (SD) | 127.4 (39.7) | 122.2 (56.3) | 0.2 |
Triglycerides, mg/dL, mean (SD) | 98.1 (53.2) | 98.6 (42.2) | 0.6 |
ESR, mm/h, mean (SD) | 22.7 (22.2) | 39.1 (36.0) | 0.038* |
Systolic blood pressure, mm Hg, mean (SD) | 151.3 (23.6) | 160.8 (30.6) | 0.2 |
Systolic blood pressure (≥180 mm Hg), mean (SD), n (%) | 10 (15) | 14 (33) | 0.030* |
Diastolic blood pressure, mm Hg, mean (SD) | 85.9 (13.7) | 91.1 (15.3) | 0.13 |
Baseline ASPECTS score <6, n (%) | 15 (27) | 20 (56) | 0.007** |
TOAST classification, n (%) | |||
Cardioembolism | 16 (25) | 12 (28) | 0.039* |
Large vessel disease | 17 (26) | 18 (42) | |
Small vessel disease | 8 (12) | 0 (0) | |
Stroke of other determined etiology | 1 (1.5) | 2 (4.7) | |
Stroke of undetermined etiology | 23 (35) | 11 (26) | |
IV rt-PA, n (%) | 34 (52) | 9 (21) | 0.001** |
EVT, n (%) | 17 (26) | 18 (42) | 0.088 |
Bridging Therapy, n (%) | 5 (7.7) | 1 (2.3) | 0.4 |
HT detected in patients, n (%) | 11 (17) | 17 (40) | 0.009** |
Baseline NIHSS score, median (IQR) | 10.0 (7.0, 15.0) | 17.0 (11.0, 20.0) | 0.003** |
NIHSS score at discharge, median (IQR) | 2.0 (0.0, 6.2) | 12.0 (7.5, 15.5) | <0.001*** |
mRS at discharge (3–6), n (%) | 31 (48) | 41 (95) | <0.001*** |
Mortality, n (%) | 0 (0) | 11 (25.5) | <0.001*** |
Legend: min, minutes; NIHSS, National Institutes of Health Stroke Scale; ASPECTS, Alberta stroke program early CT score; HT, hemorrhagic transformation; mg/dL, milligrams per deciliter; SD, standard deviation; IQR, interquartile range; mRS, modified Rankin Scale; IV rt-PA, intravenous recombinant tissue plasminogen activator; EVT, endovascular therapy; TOAST, Trial of Org 10172 in Acute Stroke Treatment; ESR, erythrocyte sedimentation rate; LDL, low-density lipoproteins; HDL, high-density lipoproteins.
1Pearson’s χ2 test; Wilcoxon rank sum test; Fisher’s exact test (*p < 0.05, **p < 0.01 and ***p < 0.001).
The mean age was 62.3 ± 11.7 years and 70% were men. At 90 days, 65 (60.1%) patients had a good functional outcome. The median baseline NIHSS score was 12 IQR (8-18). Among the etiological subtypes, large vessel atherosclerotic disease (p = 0.039) was the most common. There was no significant difference between the two groups in the time of arrival to the hospital after onset. The presence of elevated plasma glucose (196.7 ± 91.0 mg/dL; p = 0.016), stroke severity 17.0 IQR (11.0–20.0) (p = 0.003), and HT of the infarct (p = 0.009) predicted poor functional outcome.
The temporal profile of all the biomarkers given in Figure 2 showed an overall increase in the plasma levels at 12 h. In addition, sST2 showed a significant elevation at the 12-h and 24-h time points, and poor outcome was correlated at 12 h and 24 h with mean levels of 50.4 ± 51.0 ng/mL (p = 0.047) and 81.8 ± 101.3 ng/mL (p = 0.001), respectively. There were no correlations between MMP-9 and claudin-5 at 12 h with the 90-day functional outcome. Pearson product-moment correlation showed significant associations between mRS at discharge and 3 months from onset, and the sST2 levels measured at 12 h and 24 h (online suppl. Table 1; for all online suppl. material, see www.karger.com/doi/10.1159/000529512).
The temporal profile of plasma levels of MMP-9 (a), claudin-5 (b), and sST2 (c) at baseline, 12 h, and 24 h. p values are from Wilcoxon rank sum exact test, Wilcoxon rank sum test, and Fisher’s exact test (*p < 0.05, and **p < 0.01). MMP-9, matrix metalloproteinase-9; sST2, soluble serum stimulation-2; ng/mL, nanogram per milliliter; pg/mL, picogram per milliliter; mRS, modified Rankin Scale.
The temporal profile of plasma levels of MMP-9 (a), claudin-5 (b), and sST2 (c) at baseline, 12 h, and 24 h. p values are from Wilcoxon rank sum exact test, Wilcoxon rank sum test, and Fisher’s exact test (*p < 0.05, and **p < 0.01). MMP-9, matrix metalloproteinase-9; sST2, soluble serum stimulation-2; ng/mL, nanogram per milliliter; pg/mL, picogram per milliliter; mRS, modified Rankin Scale.
The ROC curve in Table 2 showed that the diagnostic accuracy of plasma sST2 levels at 24 h was 0.670 for a cut-off value of 71.8 ng/mL (specificity: 96.9%, sensitivity: 38.5%). MMP-9 levels showed a higher sensitivity for all 3 time points with an optimum cut-off value at 12-h time point (77.2 ng/mL; sensitivity: 86.1% and specificity: 44.2%). The highest sensitivity (92.3%) was attained at 24 h for a cut-off value of 50.6 ng/mL, but the AUC was only 56.6%. Claudin-5 had the lowest AUC values with low sensitivity for all 3 time points. Multiple logistic regression analysis is given in Table 3 and showed that elevated sST2 levels assessed at 24 h from onset emerged as an independent predictor of poor functional outcome at 3 months (OR: 6.44, 95% CI: 1.40–46.3, p = 0.029) by applying the cut-off value derived from the ROC curve analysis to the model.
ROC curves of the MMP-9, sST2, and claudin-5 levels at each time point
Biomarker levels at each time point . | AUC . | Cut-off value . | Sensitivity, % . | Specificity, % . | PPV, % . | NPV, % . |
---|---|---|---|---|---|---|
MMP-9 | ||||||
At baseline | 0.585 | 56.063 ng/mL | 81.8 | 39.3 | 21.4 | 55.7 |
At 12 h | 0.610 | 77.212 ng/mL | 86.1 | 44.2 | 17.9 | 48.3 |
At 24 h | 0.566 | 50.663 ng/mL | 92.3 | 30.8 | 13.0 | 55.6 |
Claudin-5 | ||||||
At baseline | 0.474 | 116.778 pg/mL | 9.1 | 98.2 | 35.2 | 25.0 |
At 12 h | 0.417 | 168.102 pg/mL | 11.1 | 96.2 | 39.0 | 33.3 |
At 24 h | 0.409 | 164.383 pg/mL | 5.1 | 96.9 | 37.0 | 50.0 |
sST2 | ||||||
At baseline | 0.593 | 11.097 ng/mL | 75.8 | 44.6 | 24.2 | 55.4 |
At 12 h | 0.619 | 36.674 ng/mL | 44.4 | 90.4 | 29.9 | 23.8 |
At 24 h | 0.670 | 71.815 ng/mL | 38.5 | 96.9 | 27.6 | 11.8 |
Biomarker levels at each time point . | AUC . | Cut-off value . | Sensitivity, % . | Specificity, % . | PPV, % . | NPV, % . |
---|---|---|---|---|---|---|
MMP-9 | ||||||
At baseline | 0.585 | 56.063 ng/mL | 81.8 | 39.3 | 21.4 | 55.7 |
At 12 h | 0.610 | 77.212 ng/mL | 86.1 | 44.2 | 17.9 | 48.3 |
At 24 h | 0.566 | 50.663 ng/mL | 92.3 | 30.8 | 13.0 | 55.6 |
Claudin-5 | ||||||
At baseline | 0.474 | 116.778 pg/mL | 9.1 | 98.2 | 35.2 | 25.0 |
At 12 h | 0.417 | 168.102 pg/mL | 11.1 | 96.2 | 39.0 | 33.3 |
At 24 h | 0.409 | 164.383 pg/mL | 5.1 | 96.9 | 37.0 | 50.0 |
sST2 | ||||||
At baseline | 0.593 | 11.097 ng/mL | 75.8 | 44.6 | 24.2 | 55.4 |
At 12 h | 0.619 | 36.674 ng/mL | 44.4 | 90.4 | 29.9 | 23.8 |
At 24 h | 0.670 | 71.815 ng/mL | 38.5 | 96.9 | 27.6 | 11.8 |
AUC, area under the ROC curve; PPV, positive predictive value; NPV, negative predictive value; MMP-9, matrix metalloproteinase-9; sST2, soluble serum stimulation-2; h, hours; pg/mL, picogram per milliliter; ng/mL, nanogram per milliliter.
Multiple logistic model for predicting functional outcome after adjusting for risk factors
Variable . | OR . | 95% CI . | p value . |
---|---|---|---|
Plasma glucose, >140 mg/dL | 2.06 | 0.73, 6.20 | 0.2 |
Systolic blood pressure, ≥180 mm Hg | 1.17 | 0.32, 4.14 | 0.8 |
Baseline ASPECTS score, <6 | 2.10 | 0.68, 6.54 | 0.2 |
HT detected in patients | 0.42 | 0.13, 1.30 | 0.13 |
Baseline NIHSS score, >16 | 1.87 | 0.62, 5.53 | 0.3 |
Large vessel atherosclerotic disease | 0.45 | 0.15, 1.27 | 0.13 |
sST2 levels at 24 h, ng/mL | 6.44 | 1.40, 46.3 | 0.029* |
Variable . | OR . | 95% CI . | p value . |
---|---|---|---|
Plasma glucose, >140 mg/dL | 2.06 | 0.73, 6.20 | 0.2 |
Systolic blood pressure, ≥180 mm Hg | 1.17 | 0.32, 4.14 | 0.8 |
Baseline ASPECTS score, <6 | 2.10 | 0.68, 6.54 | 0.2 |
HT detected in patients | 0.42 | 0.13, 1.30 | 0.13 |
Baseline NIHSS score, >16 | 1.87 | 0.62, 5.53 | 0.3 |
Large vessel atherosclerotic disease | 0.45 | 0.15, 1.27 | 0.13 |
sST2 levels at 24 h, ng/mL | 6.44 | 1.40, 46.3 | 0.029* |
sST2, soluble serum stimulation-2; ng/mL, nanogram per milliliter; mg/dL, milligram per dL; mRS, modified Rankin Scale; h, hours; NIHSS, National Institutes of Health Stroke Scale; ASPECTS, Alberta Stroke Program Early CT Score; HT, hemorrhagic transformation; OR, odds ratio; CI, confidence interval.
*p < 0.05.
Discussion
Our study demonstrated the prognostic impact of circulating sST2 levels measured within 24 h from stroke onset predicting short-term functional outcomes in AIS. We evaluated plasma levels of MMP-9, sST2, and claudin-5 using serial measurements from onset to observe the dynamic changes from baseline to 24 h.
Of the three markers, sST2 showed a gradual increase in its concentration across the 3 time points in the poor outcome group with significantly elevated levels at 12 h and 24 h from the onset. This temporal profile was consistent with the previous findings of its increased activity beyond 24 h from onset indicating its expression during the inflammatory phase of ischemia [11].
The biomarker sST2 independently predicted poor functional outcomes at 90 days after adjusting for covariates. The probability of an unfavorable outcome was 6 times higher when the cut-off value of sST2 levels at 24 h was applied to the model yielding better discriminative capacity and high specificity.
Our previous study had shown that MMP-9 was positively correlated with baseline stroke severity and claudin-5 was a predictor of HT, both of which are known risk factors of functional outcome [16]. Hence, we sought to determine their roles in the outcome of AIS as well. We found that there was no significant association of both these markers with the outcome, although a trend toward elevation was observed at the 12-h time point in the patients with poor outcomes. Furthermore, the discriminatory capacity of claudin-5 was low, thereby affecting the reliability of the marker.
Experimental studies have indicated the role of neuroinflammatory markers in the disruption of the BBB [2, 17, 18]. Elevated levels of sST2 have been previously reported as a prognostic indicator of AIS in a few studies [3, 4, 6]. One study showed the highest tertile of sST2 levels above a cut-off level of 22.96 ng/mL, which was measured at a median time of 19 h from the onset, and predicted poor outcome and all-cause mortality at 90 days [3]. A similar correlation was reported for an earlier time point by Wolcott and colleagues, whereby sST2 >44.6 ng/mL was independently associated with outcome and mortality in AIS [6]. These baseline measurements were in line with our findings; however, we were not able to report a significant association with the functional outcome in our population. Another recent study that looked into the relationship of sST2 levels in AIS and cardiac patients reported a significant correlation between the levels of sST2 evaluated at 24 h from admission with all-cause mortality at 3 months, but relatively lower levels at baseline and in the subsequent time points were observed in the AIS group [4]. Contrary to these findings, in a multi-marker study conducted by Dieplinger and colleagues, sST2 showed no correlation with functional outcomes [19]. Nevertheless, the heterogeneity in these results may be attributed to the sampling done at different time points of admission rather than at specific time points from the onset of stroke. This was evident in the dynamics observed in the temporal distribution of each marker in our study. sST2 levels were more pronounced during the late phase of acute stroke when inflammatory mechanisms are activated [12], but both MMP-9 and claudin-5 were found to be expressed earlier during the BBB dysfunction owing to their response to oxidative stress [10]. sST2 is a receptor of the immunomodulatory cytokine IL-33 and is a known prognostic marker for cardiac diseases as the IL-33/sST2 system was found to respond to cardiac stress [20]. However, recent studies have also implicated its key role in various central nervous diseases. Whereas IL-33 functions as neuroprotective, increased levels of circulating sST2 inhibit IL-33 and exacerbate the pro-inflammatory processes [21]. The opposing functions of IL-33 and sST2 are necessary for modulating the neuroprotective effects postischemia and remodeling the extracellular matrix that may influence the outcome after stroke [11]. This relationship is evident in a study that showed the presence of lower IL-33 levels in patients with poor outcomes in ischemic stroke [22].
The strengths of our study included meticulous sample collection at specific time points from the onset of stroke. Serial evaluation of biomarkers was taken from stroke onset rather than in the hours following admission to improve the accuracy of the outcome prediction by maintaining uniformity among samples. Few studies that looked into the association of sST2 with AIS assessed the plasma levels at baseline alone. Our study also has limitations as it follows a monocentric design, and the small sample size may affect the generalizability of our findings across other ethnic groups. We had excluded a significant number of patients based on the eligibility criteria and as these data were not collected, we were unable to report the difference between the included and excluded patients. This may have led to a selection bias in our findings. The relationship of these biomarkers may need to be further investigated in a larger sample population to establish their clinical applicability.
Conclusion
sST2 may be considered a prognostic marker for predicting short-term outcomes in AIS patients.
Statement of Ethics
This study protocol was reviewed and approved by the Institutional Ethics Committee of SCTIMST (approval number - 1295). A written informed consent was received from each patient at the time of enrolment.
Conflict of Interest Statement
The authors have no conflicts of interest to declare.
Funding Sources
This project was funded by the Technology Development Fund (TDF) under the intramural scheme of SCTIMST (Project ID IRC064).
Author Contributions
P.N.S. and S.K. were involved in the conceptualization and reviewing and drafting of the manuscript of the study. P.N.S and S.E.S were involved in project administration, supervision, validation, and funding acquisition. Data curation was carried out by S.K., and P.N.S. Methodology and formal analysis were carried out by S.K., P.N.S., G.S., D.D., S.G., and M.U.K.
Data Availability Statement
All data generated or analyzed during this study are included in this article and in its supplementary material files. Further inquiries can be directed to the corresponding author.