Role of Blood Biomarkers in Differentiating Ischemic Stroke and Intracerebral Hemorrhage
Correspondence Address: Source of Support: None, Conflict of Interest: None DOI: 10.4103/0028-3886.293467
Source of Support: None, Conflict of Interest: None
Keywords: Biomarkers, BNP, GFAP, ICH, Ischemic stroke, NR2, S100
Stroke accounts for the second leading cause of death and disability around the world. Early diagnosis of stroke is imperative to provide timely and appropriate care., Although the diagnosis of stroke is clinical, recognizing and categorizing the type of stroke is essential to provide exact therapy., About 80% of strokes are ischemic and 15%–20% are hemorrhagic, with different therapeutic and prognostic implications. Differentiating stroke subtype is a critical step when planning therapeutic goals in the initial period as specific management and treatment protocols are recommended for stroke subtypes including thrombolysis, blood pressure reduction etc. Although imaging is the mainstay of differentiating ischemic stroke (IS) from intracerebral hemorrhage (ICH), these facilities are not available in many places. In a stroke, the biomarker work is based on the pathophysiological aspects of ischemic tissue damage and the possibility that a particular protein will be high in serum after tissue damage. There could be markers of tissue damage, inflammation, endothelium or hemostasis. High throughput screening of many molecules is another way of looking for a biomarker. Other methods include genome or proteome methods and RNA expression.,,,, However, limitations to the use of biomarkers stems from the fact that tissue damage may happen over time and blood brain barrier may impede the release of these molecules into the blood stream.
Recent past has seen an increasing interest in this field with various groups around the world aiming at discovering the ideal biomarker for a particular disease process. Some proteins have been promising, yet validation is poor and there is a lack of sensitivity and specificity. Although studies have looked into various facets of stroke including diagnosis, severity, outcomes, etiology, and correlation with biomarkers like astroglial protein S100B, glial fibrillary acidic protein (GFAP), neuron specific enolase (NSE), vascular cell adhesion molecule (VCAM1), inter-cellular adhesion molecule ICAM1, N-methly-d-aspartate (NMDA) receptor antibodies and matrix metalloproteinases (MMP) and acute thrombosis molecules like D-dimer, fibrinogen, and von-Willebrand factor (vWF); none has be found to be absolute and specific, although data is encouraging. They may not be specific to a process, the blood brain barrier (BBB) restricts release of these biomarkers into the systemic circulation, and therefore may not correlate with stroke severity. The present study aimed at assessing a set of biomarkers to help differentiate IS from ICH.
This is a prospective study carried out at the Department of Neurology, All India Institute of Medical Sciences, New Delhi, India. The study was approved by the institute ethics committee. Informed consent was taken from each patient or his legally authorized representative. The primary objective of this study was to compare levels of individual blood markers S100 B, glial fibrillary acidic protein (GFAP), N-methyl-D Aspartate receptor antibody (NR2) subunit, brain natriuretic peptide (BNP), and interleukin 6 (IL6) between IS and ICH. The secondary objective was to see if any particular biomarker correlated with specific stroke etiology as classified using Trial of Org10172 in Acute Stroke Treatment (TOAST) criteria. Based on previously published observations, with an alpha error of 5% and a 90% power, a sample size of 250 patients was planned. This estimate includes the fact that in any stroke cohort, ischemic stroke would be about 75%–80% and ICH about 20%–25%.
Consecutive patients with acute stroke within 24 hours of presentation were recruited in the study. The stroke was confirmed by clinical features and imaging including computed tomography (CT), CT angiogram or magnetic resonance imaging (MRI), MR angiogram. Patients were divided into ischemic stroke (IS) and intracerebral hemorrhage (ICH) based on imaging. Details of stroke onset time, demographics, stroke risk factors, imaging details, baseline stroke severity (using National Institutes of Health Stroke Scale (NIHSS) scale), therapy received, and etiology of stroke were recorded. Outcomes were assessed using modified Rankin score at three months. Patients with stroke >24 hours from onset, ICH due to etiology other than hypertension, active infection, active systemic inflammatory disease, current immunosuppressive therapy, and neurological disease other than stroke including dementia, tumor, and infection were excluded.
Each patient was recruited as close to the index event upon presentation to the emergency and a blood sample was drawn for biomarker analysis. The sample was immediately centrifuged at 3000 rpm for 10 min and serum separated and stored at -80 C, till analysis. IL6 was measured by a solid phase, competitive chemiluminescent enzyme immunoassay method using the Immulite, Siemens machine and BNP, S100, GFAP, and NMDA receptor antibody nR2 subunits were measured using the ELISA technique with commercial kits. The data was entered in the microsoft excel format and analyzed using Stata version 14.2. Continuous variables were assessed using Student's t test and categorical by Fisher's exact test. Data is expressed as mean SD or median (IQR) where appropriate. ROC curves were drawn for each biomarker and sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) measured against the clinical and imaging based gold standard diagnosis. Sensitivity, specificity and predictive values were calculated along with 95% confidence intervals. Association of biomarker levels with different etiologies was assessed using one-way analysis of variance (ANOVA).
250 acute stroke patients were recruited into the study. Among these, 187 were IS and 63 patients had ICH. Baseline variables are presented in [Table 1]. The mean age of IS patients was 55.70 ± 14.78 years and ICH was 59.12 ± 15.10 years. There were 64.71% (121/187) males among IS and 60.3% (39/63) among ICH group. The median NIHSS was higher among patients with ICH and higher number of patients in the IS group were hypertensives. Coronary artery disease and atrial fibrillation were significantly higher among patients with IS.
The observed values are presented in [Table 2] and [Table 3]. [Table 2] depicts the mean and median values of biomarkers among the two groups. [Table 3] depicts a comparison between the two groups. Patients with extreme values were considered outliers and were removed from analysis for more homogenous results (S100 four, NR2 one, IL6 4). S100 values was higher in patients with ICH compared with IS (8 pg/ml versus 4.2 pg/ml respectively, P = 0.003) and IL6 was higher in patients with IS compared to ICH (12.9 pg/ml vs 8.76 pg/ml, P = 0.02). No significant differences were found among the other biomarkers. The area under the curve (AUC) for all biomarkers is outlined in [Table 4]. AUC for S100 was 65% (95% CI 0.58–0.73, [Figure 1]). At a value of 5.5 pg/ml and above, S 100 had a sensitivity of 61.9% (95% CI 48.8–73.9), specificity of 61.2% (95% CI 53.70–68.3), PPV of 35.5% (95% CI 26.6–45.10), and a NPV of 82.4% (95% CI 74.9–88.4) to diagnose ICH. The area under the curve (AUC) of IL6 was 59% (95% CI 0.53–0.61, [Figure 1]). At a value of less than 11 pg/ml, IL6 had a sensitivity of 55.7% (95% CI 42.4–68.5), specificity of 56.2% (95% CI 48.7–63.5), PPV of 29.6% (95% CI 21.4–38.8) and NPV of 79.4 (95% CI 71.4–86.0) to diagnose ICH. We assessed stratified values of S100 and IL6 to assess for the discriminatory ability to diagnose ICH. Values higher than 5 pg/ml were significantly more in patients with ICH. Similarly, for IL6, value higher than 10 pg/ml was more among patients with IS. We combined the results of S100 and IL6, to assess if together they have better discriminatory ability to diagnose ICH when compared with IS. Levels were considered normal if S100 <5.5 pg/ml and IL6 >10 pg/ml or abnormal if any one value was abnormal i.e: S100 >5.5 pg/ml and/or IL6 <10pg/ml. The abnormal value was seen among 91.38% patients with ICH compared with 63.74% patients with IS (P = 0.001) with a sensitivity of 91.4%, specificity of 36.3%, PPV of 32.7% and a higher of 92.5%.
We also assessed biomarker values along different time points of sample collection from index event although individual patients were not sampled periodically over time. No statistical differences were observed among the biomarkers with respect to time. [Figure 2]a and b] Using TOAST classification, levels of biomarkers were analyzed in various etiological subtypes. No statistically significant differences were observed among various biomarkers and TOAST subtypes.
A blood biomarker should ideally be sensitive, specific, highly accurate, reproducible, have good positive and negative predictive values, be easy to interpret by clinicians, and cost effective.,, There has been an interest in developing biomarkers to differentiate stroke from mimics, types of stroke, establishing etiology, and outcomes in general. A panel of proteins, S100 B, VWF, MMP9, VCAM or S100B or VWF, MMP9, BDNF, and MCP-1 can differentiate IS from controls with a high sensitivity and specificity., In a systematic review of blood biomarkers in the diagnosis of ischemic stroke, the authors observed that although all showed a high sensitivity or specificity, there were significant limitations in the design and reporting of all studies.
In a recently updated evidence review, the authors concluded that there is still not enough evidence to suggest that any newer biomarkers could differentiate IS from other causes. In a large multicentric study (BRAIN, The Biomarker Rapid Assessment In Ischemic Injury Study), 1146 patients with symptoms suggestive of stroke were recruited from 17 centers. Blood was analyzed for MMP9, BNP, D-Dimer, and protein S100. A diagnostic tool incorporating the values of matrix metalloproteinase 9, brain natriuretic factor, D-dimer, and S-100 into a composite score was sensitive for acute cerebral ischemia. The multivariate model demonstrated modest discriminative capabilities with an area under the receiver operating characteristic curve of 0.76 for hemorrhagic stroke and 0.69 for all stroke (likelihood test P < 0.001). When the threshold for the logistic model was set at the first quartile, this resulted in a sensitivity of 86% for detecting all stroke and a sensitivity of 94% for detecting hemorrhagic stroke. However, the specificity was only 36%. Moreover, results were reproducible in a separate cohort tested on a point-of-care platform. The authors concluded that these results suggest that a biomarker panel may add valuable and time-sensitive diagnostic information in the early evaluation of stroke. Such an approach is feasible on a point-of-care platform.
Few studies have attempted differentiation of IS from ICH. In our current study, only S100 and IL6 had significant discriminatory ability to diagnose ICH, albeit with a moderate sensitivity and specificity and a higher negative predictive value. In a series of 135 patients (IS = 93, ICH = 42) assessed within 6 hours of stroke onset GFAP (glial fibrillary acidic protein) was detectable in the serum of 39 patients (34/42, 81% with ICH and 5/93, 5% of IS). Serum GFAP was substantially raised in patients with ICH compared with IS and was able to differentiate IS from HS within six hours of onset with 79% sensitivity, 98% specificity, PPV 94% and NPV of 91%, P < 0.001 with a ROC cut-off of 2.9 ng/L levels. Another study of 205 patients from 14 centers across Germany and Switzerland aimed at studying GFAP as a candidate biomarker among ICH patients, within 4.5 hours of onset of stroke. Among 39 ICH and 163 IS patients, GFAP concentrations were increased in patients with ICH compared with patients with ischemic stroke [median (interquartile range) 1.91 μg/L (0.41–17.66) vs 0.08 μg/L (0.02– 0.14), P < 0.001]. Diagnostic accuracy of GFAP for differentiating ICH from I Sand stroke mimic was high [area under the curve 0.915 (95% CI0.847– 0.982), P < 0.001]. A GFAP cutoff of 0.29 μg/L provided diagnostic sensitivity of 84.2% and diagnostic specificity of 96.3% for differentiating ICH from IS and stroke mimic. In the present study, we found a trend towards a higher GFAP levels, among patients with ICH, although it could not reach a statistical significance (P = 0.09) even though the proportion of patients with ICH (n = 63) versus IS (n = 189) were similar in the published study above. A larger dataset may be required to validate this finding further. It could also be related to timings of the sample collection as not all patients could be sampled closest to the stroke time considering there varied times of presentations to the hospital after the index event. Timing from stroke onset has huge relevance in case we wish to triage the patients for specific treatment and blood biomarkers could vary with time duration.
Using CRP, D-dimer, sRAGE, MMP9, S100B, BNP, NT-3, caspase 3, and chimerin –II, as biomarkers in 915 patients, the authors in this study found significantly high levels of S100 and reduced levels of sRAGE in patients with ICH. The complete protocol was achieved in 915 patients (776 IS, 139 ICH). Among blood samples obtained <6 h from symptoms onset (n = 337), S100B levels were increased in ICH (107.58 vs 58.70 pg/mL; P < 0.001), whereas sRAGE levels were decreased (0.77 vs 1.0 2ng/mL; P = 0.009) as compared to IS. In this subset of patients S100 B (OR 3.97 95% CI 1.82–8.68; P = 0.001) and sRAGE (OR 0.22 95% CI 0.10–0.52;
P < 0.001) were independently associated with ICH. A regression tree was created by CART method showing a good classification ability (AUC = 0.762). Similar results were found for samples obtained within three hours. The authors concluded that a combination of biomarkers including those of the S100B/RAGE pathway seems promising to achieve a rapid biochemical diagnosis of IS versus ICH in the first hours from symptoms onset. We also found S100 to be a likely promising marker with statistically higher levels among patients with ICH compared with IS subjects. Upon stratification, values above 5ng/ml were consistently high among patients with ICH versus IS. This biomarker denotes neuronal injury and is therefore likely to be increased in patients with ICH, which leads to more shearing damage to brain tissue. However, the same results have not been replicated in another study where 97 prospective stroke patients (86% IS and 14% ICH) in a multicenter design were evaluated with blood levels of S100B, NSE, GFAP and coagulation biomarker, activated protein C inhibitor complex (APC-PCI) within 24 hours of onset. There were no differences in S100B (P = 0.13) and NSE (P = 0.67) levels between patients with ischemic stroke or ICH. However, GFAP levels were significantly higher in ICH patients (P = 0.0057). APC-PCI levels were higher in larger ischemic strokes (P = 0.020). The combination of GFAP and APC-PCI levels, in patients with NIHSS score more than 3, had a sensitivity and negative predictive value of 100% for ICH (P = 0.0052). The authors concluded that blood levels of biomarkers GFAP and APC-PCI, prior to neuroimaging, may rule out ICH in a mixed stroke population. We also explored if a combination of S100 and IL6 would be more diagnostic to differentiate ICH from IS. It was observed that any one abnormal value (based on combinations) increased the negative predictive value to rule out ICH. In recent stroke-chip study, using a set of 21 biomarkers, the authors could only find N-Terminal Pro-B-Type natriuretic peptide to be an independent parameter to differentiate IS from ICH. Although this was a large multicentric study, it failed to suggest many potential biomarkers as promising to differentiate stroke from mimics and IS from ICH.
In a recent meta analysis, although activated rotein C- protein C inhibitor complex (APC-PCI), glial fibrillary acidic protein (GFAP) and a panel of APC-PCI &GFAP and retinol binding protein 4 (RBP4) &GFAP were found to have high sensitivity and specificity for differentiating the two stroke types, the authors felt that the data was yet not robust to suggest its routine clinical use. Preliminary studies of NR2 peptides (degradation products of NR2 receptors of NMDA) as biomarkers supports there use in diagnosing stroke. Following stroke, NR2 peptide levels rise. IgG antibodies to the fragments of the NMDA receptor have been recognized and can be detected in blood and is a potential biomarker.,, However, NMDA-R antibodies have also been associated with hypertension, atherosclerosis, prior stroke, epilepsy, and encephalitis. Thus, the specificity of these antibodies is uncertain. We could not however, replicate this finding in our study. In a recent study, vascular endothelial growth factor (VEGF) was observed to correlate with three months outcome and stroke severity. The authors used a clinical biomarker module by incorporating biomarker levels of VEGF and clinical parameters to show a good prediction of three months outcome. Imaging characteristics of patients may also reveal an underlying etiology.
The present study did not find a significant difference in biomarker levels based on underlying etiology. It could potentially be related to a smaller sample size or inability to definitely characterize patients in a defined etiology. In a recent observation, the authors found multiple vascular territory acute infarcts to be related to many causes although cardioembolism remained the commonest. Higher levels of NT-proBNP have been observed in patients with cardioembolic stroke.
Our study has limitations. The samples were collected from patients up to 24 hours from the index event. Although this is more realistic and pragmatic as access to ideal ambulance services are not available to many patients and pre-stroke notification processes may be far from ideal, yet, the ideal biomarker should be assessed closest to stroke onset for its stratification and timely referral. With the current treatment for endovascular interventions being extended to 24 hours, the utility may be of value. A larger sample size may require validation and methods to detect these biomarkers need to be made accessible using a rapid assay like an automated small device using strip technology for wider application.
In conclusion, S100 and IL6 are potential biomarkers for further study and validation. Studies for newer biomarkers with a higher discriminatory ability or a combination of biomarkers are required in the future for diagnostic use.
Acknowledgement Grant Support
Funded by Intramural research grant by the All India Institute of Medical Sciences, New Delhi, India.
Financial support and sponsorship
Grant support under the acknowledgement section above.
Conflicts of interest
There are no conflicts of interest.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4]