S. no
Study
Sample size
Protein profile assayed
Methodology
Duration of stroke
Conclusion
Limitations of study
Sensitivity
Specificity
1.
Allard (2004) (Switzerland) [7]
26—IS
19—HS
Validated
11—IS
10—HS
Plasma proteins
SELDI, LC-ESI-MS, ELISA
6 h (range 40 min to 3 days)
Apo C-I and Apo C-III, first plasmatic biomarkers to distinguish IS and HS in a small no. of patients
Small sample size in discovery and validation phases
Apo C-I
94%
Apo C-III
94%
Apo C-I
73%
Apo C-III
87%
2.
Montaner (2012) (Spain) [23]
776—IS
139—ICH
CRP, D-dimer, sRAGE, MMP9, S100B, BNP, NT-3, caspase-3, chimerin-II, secretagogin, Cerebellin, NPY (Plasma)
ELISA
Within 24 h; (<6 h; <3 h)
S100B and sRAGE as a rapid blood test might help to distinguish IS and HS
Validation not performed and low sensitivity of biomarkers reported, clinically relevant control group not taken
22.7%
80.2%
3.
Sharma (2014) (USA) [27]
57—IS
32—ICH
262 Serum Biomarkers
ELISA
Within 24 h
5 biomarkers (eotaxin, EGFR, S100A12, TIMP-4, and prolactin) distinguished IS and ICH (C = 0.82)
Sample size not large enough to perform external validation, sensitivity and specificity of the panel to distinguish IS and HS not reported
–
–
4.
Walsh (2016) (USA) [9]
14—IS
23—HS
Apo A-I, Apo C-I, Apo C-III, MMP-3, MMP-9, and paraoxonase-1 (Plasma)
Multiplex Assays, ELISA
<12 h
Apo A-I and paraoxonase-1 levels differed between IS and HS cases
Small sample size, validation phase not performed
–
–
5.
Lopez (2012) (USA) [8]
54—IS
26—HS
9 SerumApolipoproteins
Multiplex SRM Assay (MS)
<1 week
apo C-III and apo A-I differentiated IS and HS
Validation required in large samples, samples in first few hours required
–
–
6.
Kavalci (2011) (Turkey) [24]
71—IS
29—HS
Validated
100—Stroke
4 Plasma biomarkers
Triage stroke panel—a biochemical multimarker assay
Within 24 h
A combination of BNP, D-dimer, MMP9, and S100b plasma biomarkers can differ IS and HS
Study not adequately powered, sensitivity and specificity of the panel not high, larger validation cohort required
BNP
65.5%
D-dimer
58.6%
MMP-9
65.5%
S100B
13.8%
BNP
60.6%
D-dimer
59.2%
MMP-9
66.2%
S100B
98.6%
7.
Roudbary (2011) (Iran) [22]
16—IS
16—HS
hs-CRP (serum)
Immunonephelometric method
Within 24 h
hs-CRP level is increased in patients with IS but not in HS
Less sample size, Validation not done, clinically relevant control group not taken
–
–
8.
Llombart (2016) (Spain) [19]
Discovery
36—IS
10—HS
First rep.
16—IS
16—HS
Second rep.
38—IS
28—HS
RBP4, GFAP (Plasma)
Multiplex Sandwich ELISA
<6 h
RBP4 and GFAP useful as diagnostic biomarkers to differentiate IS and ICH
Results obtained in the discovery phase were not corrected for multiple testing; replication cohorts are small
RBP4
68.4%
GFAP
32%
RBP4
84%
GFAP
100%
9.
Foerch (2012) (Germany) [14]
163—IS
39—HS
Plasma GFAP
Electrochemiluminometric immunoassay
<4.5 h
GFAP test performed within 4.5 h of symptom onset is a reliable tool for thedifferentiation between ICH and IS
Validation phase not performed, only a phase-1 study, less number of HS patients recruited
84.2%
96.3%
10.
Foerch (2006) (Germany) [13]
93—IS
42—HS
Serum GFAP
Elecsys
<6 h
GFAP raised in ICH as compared to IS in first 6 h
Determined GFAP at time point of hospital admission, not in prehospital setting; validation with same cutoff required in other populations
79%
98%
11.
Dvorak (2009) (Germany) [15]
45—IS
18—HS
Serum GFAP
ELISA
2–48 h
GFAP was reported to be higher in ICH than in IS with best time window between 2 and 6 h
Small sample size; clinically relevant control group not taken
70%
100%
12.
Unden (2009) (Sweden) [16]
83—IS
14—HS
S100B, NSE, GFAP, APC-PCI (Serum)
ELISA
<24 h
Admission GFAP and APC-PCI levels may rule out ICH
Small sample size; validation with same cutoff required in other populations; clinically relevant control group not taken, less number of HS patients
APC-PCI 96%
GFAP 79%
Panel (GFAP, APC-PCI) 71%
APC-PCI 42%
GFAP 64%
Panel (GFAP, APC-PCI) 73%
13.
Xiong (2015) (China) [17]
65—IS
43—HS
Serum GFAP
ELISA
~2–6 h
GFAP test within 2–6 h after stroke onset could be used to differentiate ICH and IS
Small Sample size; clinically relevant control group not taken; influence of other parameters on GFAP level not considered
86%
76.9%
14.
Ren (2016) (China) [18]
79—IS
45—ICH
Serum UCH-L1 and GFAP
Sandwich ELISA
Within 4.5 h
GFAP differentiated between IS and ICH
Modest sample size; validation phase not performed
GFAP 61%
GFAP 96%
15.
Rainer (2007) (China) [26]
118—IS
35—HS
Serum S100 and Plasma DNA
qRT-PCR, ELISA
<24 h
A combination of S100 and Plasma DNA differentiated ICH from IS
Less sample size, validation phase not performed
S100
47%
Plasma DNA
31%
S100
81%
Plasma DNA
83%
16.
Kim (2010) (South Korea) [10]
89—IS
11—HS
BNP, D-dimer, MMP-9, S100B (Plasma)
Fluorescence Immunoassay
6 h (range: 0–120 h)
Only BNP distinguished between AIS and HS
Less sample size in HS, validation not performed
–
–
17.
Luger (2017) (Germany) [20]
146—IS
45—ICH
Serum GFAP
Electrochemiluminometric immunoassay
Within 6 h
GFAP differentiated ICH from IS
Less number of ICH cases recruited, no validation phase
77.8%
94.2%
18.
Zhou (2016) (China) [21]
71—IS
46—ICH
Plasma S100B
Electrochemiluminescence immunoassay
Within 6 h
S100B distinguished between IS and ICH
Small sample size, no validation phase performed
95.7%
70.4%
19.
Katsanos (2017) (Greece) [11]
121—IS
34—ICH
Plasma GFAP
ELISA
Within 6 h
Plasma GFAP distinguished ICH from IS with optimum diagnostic yield
Small sample size, no validation phase
91%
97%
20.
Rozanski (2017) (Germany) [12]
49—IS
25—ICH
Plasma GFAP
ELISA
62.5 min (36–139 min)
GFAP levels >0.29 ng/ml were seen only in ICH, thus confirming the diagnosis of ICH during prehospital care
Small sample size, IS patients significantly older than ICH patients, validation phase not performed, poor sensitivity
36%
100%
21.
Bustamante (2017) (Spain) [25]
941—IS
174—HS
21 protein biomarkers
ELISA
Within 6 h
Only NT-proBNP and endostatin differentiated IS from HS with moderate predictive accuracy of 80.6%
Low reproducibility between interim and final analysis, poor sensitivity
NT-pro BNP
44.8%
Endostatin
18.8%
NT-pro BNP
74.9%
Endostatin
90.8%
2.2 Blood Biomarkers for the Differentiation of Ischemic Stroke from Stroke Mimics
2.2.1 Individual Biomarkers
In a small sample of 34 IS and 29 mimics, Glickmann et al. in 2011 [28] analyzed five serum based biomarkers, namely, CRP, MMP-9, S100B, BNP, and D-dimer for their potential to diagnose IS from stroke mimics. They observed a model of C-Reactive Protein and National Institutes of Health Stroke Scale (CRP and NIHSS) to be highly predictive of IS with a discrimination capacity of C = 0.95. The levels of CRP biomarker were significantly higher in IS patients as compared to stroke mimics (IS: 37.6 ± 33.1 vs. mimics: 9.7 ± 11, p < 0.001). MMP-9 and S100B were found to be moderately predictive of IS when combined with NIHSS in a bivariate model with a discrimination capacity of C = 0.92 for MMP-9 and C = 0.87 for S100B. The median time from symptom onset to blood sampling was 5 h (interquartile range: 3.5–8 h). The levels of S100B were also assessed by Gonzalez-Garcia et al. in 2012 [29] along with neuron-specific enolase (NSE) to diagnose total stroke from mimics + transient ischemic attack (TIA) within 48 h of symptom onset. In a sample of 61 stroke (IS = 44, HS = 17) and 11 TIA + mimics, NSE had a sensitivity of 53% and specificity of 64% at a cutoff value of 14 μg/l, while S100B had a sensitivity of 55% and specificity of 64% at a cutoff value of 130 ng/l to differentiate total stroke from TIA + mimics. They concluded that neither NSE nor S100B improved the diagnosis of acute stroke. Another study by An et al. in 2013 [30] assayed the levels of MMP-9 and S100B along with NSE, VSNL-1, hFABP, and Ngb, GFAP, MMP-9, IL-6 and TNF-α to diagnose IS within 24 h of symptom onset. Only IL-6 (IS: 4.0 [0.8–12.3]; mimic: 1.2 [0.0–2.4]; p < 0.001), S100B (IS: 30.4 [0.0–115.2]; mimic: 2.3 [0.0–20.6]; p < 0.001), and MMP-9 (IS: 63.3 [29.7–122.8]; mimic: 33.8 [15.4–60.8]; p < 0.001) were found to be significantly elevated in 188 IS as compared to 90 stroke mimics.
A study published by Airas et al. in 2008 [31] observed the levels of vascular adhesion protein-1 (VAP-1) to be significantly elevated in 20 IS cases as compared to 20 mimics (IS: 652 ± 224 ng/ml, Mimics: 542 ± 104 ng/ml; p < 0.05) within 6 h of symptom onset.
Doehner et al. in 2012 [32] studied the plasma levels of neuropeptides proenkephalin A (PENK-A) and protachykinin (PTA) in 124 IS, 16 TIA, and 49 mimic cases. PENK-A concentration was significantly elevated in patients with stroke (median [IQR]): 123.8 pmol/l [93–160.5]) compared to patients with TIA (114.5 pmol/l [85.3–138.8]) and with mimics (102.8 pmol/l [76.4–137.6]; both groups vs. stroke p < 0.05). However, no significant difference was observed in PTA levels between these groups. Ahn et al. in 2011 [33] compared the usefulness of albumin-adjusted ischemia-modified albumin index (IMA index) to the ischemia modified albumin (IMA) in early detection of IS from mimics in a small sample of 28 IS and 24 mimics. IMA index was found to be more sensitive (sensitivity: 95.8%, specificity: 96.4, cutoff: 91.4 U/ml) than conventional IMA (sensitivity: 87.5%, specificity: 89.3%, cutoff: 98 U/ml) to diagnose IS from mimics within 6 h of symptom onset. Meng et al. in 2011 [34] assayed several plasma biomarkers including antithrombin III (AT III), thrombin–antithrombin III (TAT), fibrinogen, D-dimer, and high-sensitivity C-reactive protein (hsCRP) to differentiate 152 IS patients from 46 stroke mimics within 4.5 h of symptom onset. Only AT-III (sensitivity: 97.37, specificity: 93.62, cutoff: 210%) and fibrinogen (sensitivity: 96.05, specificity: 82.61, cutoff: 4 g/l) were able to differentiate IS from mimics with adequate sensitivities and specificities. Dambinova et al. in 2012 [35] tested the potential of NR2 peptide to differentiate IS from stroke mimics in a sample of 101 IS and 91 mimics within 72 h of symptom onset. NR2 peptide was found to have a sensitivity of 92% and specificity of 96% to differentiate IS from stroke mimics at a cutoff value of 1.0 μg/l. Dassan et al. in 2012 [36] found limited clinical utility of serum vascular endothelial growth factor (VEGF) in being able to differentiate IS from stroke mimics within 24 h of symptom onset. In a small sample of 29 IS and 15 mimics, VEGF was found to have a modest sensitivity of 69% and specificity of 73% to differentiate IS from mimics at a cutoff value of 1026 pg/ml. Wendt et al. in 2015 [37] did not find copeptin to be an appropriate biomarker to discriminate between stroke and mimics. No statistically significant difference was observed in the copeptin levels between stroke and mimics (11.5 (5.3–29.3) vs. 8.2 (3.3–32.8), p = 0.15).
2.2.2 Panel of Biomarkers
Laskowitz et al. in 2009 [38] published a multicenter study at 17 centers to test the diagnostic performance of a biomarker panel comprising of MMP-9, BNP, D-dimer and S100B to diagnose IS. In 293 IS and 361 stroke mimic cases, they observed a sensitivity of 85% and a specificity of 34% for the 25th percentile while a sensitivity of 36% and a specificity of 84% for the 75th percentile of the biomarker panel to diagnose IS from stroke mimics within 24 h of symptom onset. The performance of this panel was also assessed by a different group in a study published by Sibon et al. in 2009 [39]. This group compared the accuracy of the Triage stroke panel consisting of D-dimer, BNP, MMP-9 and S100B to the triaging nurse for diagnosis of stroke from mimics. In 126 stroke (85 IS, 33 TIA, 13 HS) and 65 mimic cases, the biomarker panel was found to have an equivalent sensitivity of 92.9% and a specificity of 23.8% to differentiate and diagnose total stroke from stroke mimics to that of the nurse. A few years later, Knauer et al. in 2012 [40] tested the potential of this panel of biomarkers and they did not recommend the use of BNP, D-dimer, MMP-9, and S100B as a panel to distinguish IS from mimics. Receiver operating characteristic (ROC) curve analysis observed low discriminating power of the panel of biomarkers with an area under the curve (AUC) of 0.59. The model had a high sensitivity of 92% but a low specificity of 14% while a low sensitivity of 14% and a high specificity of 86% to differentiate IS from mimics within 6 h of symptom onset at cutoff values of 1.3 and 5.9 ng/ml respectively.
Montaner et al. in 2011 [41] found a panel of 6 plasma biomarkers namely caspase-3, D-dimer, sRAGE, chimerin, secretagogin and MMP-9 which differentiated total stroke (IS = 776 + HS = 139) from 90 stroke mimics. Within 24 h of symptom onset, the model had a sensitivity of 17% and specificity of 98% in the first quartile (cutoff value of 0.87) and a sensitivity of 82% and specificity of 59% in the last quartile (cutoff value of 0.97) to differentiate total stroke from mimics. In the blood samples obtained within 3 h, the model had a sensitivity of 87% and specificity of 55% in the first quartile (cutoff value of 0.49) while a sensitivity of 28% and specificity of 99% in the last quartile (cutoff value of 0.87) for differentiating total stroke from mimics. When the levels of secretagogin and chimerin were low and the levels of caspase-3, D-dimer, sRAGE, and MMP-9 were high, the model had a 100% probability of predicting stroke.
Sharma et al. in 2014 [27] assayed a total of 262 serum biomarkers in 57 IS patients and 37 stroke mimics and identified a panel of five biomarkers including eotaxin, EGFR, S100A12, TIMP-4, and prolactin which differentiated IS patients from mimics with a discriminative capacity of C = 0.92 within 24 h of the onset of event.
Studies determining blood-based biomarkers for the differentiation of ischemic stroke from stroke mimics
S. no | Study | Sample size | Protein profile | Methodology | Duration of stroke | Conclusion | Limitations of the study | Sensitivity | Specificity |
|---|---|---|---|---|---|---|---|---|---|
1. | Sharma (2014) (USA) [27] | 57—IS 37—Mimics | 262 Serum Biomarkers | ELISA | Within 24 h | Five biomarkers (eotaxin, EGFR, S100A12, TIMP-4, and prolactin) differentiated stroke from stroke mimics | Sample size not large enough to perform external validation | 90% | 84% |
2. | Glickman (2011) (USA) [28] | 34—IS 29—Mimics | BNP, CRP, D-dimer, MMP9, S100B (Serum) | Assay | 5 h (3.5–8 h) | CRP has high discriminating capacity and MMP-9 and S100B have moderate discriminating capacity to distinguish IS from mimics | Less sample size, validation phase not done | – | – |
3. | Montaner (2011) (Spain) [41] | 776—IS 90—Mimics | CRP, D-dimer, sRAGE, MMP-9, S100B, BNP, caspase-3, neurotrophin-3, chimerin, secretagogin | Sandwich ELISA | Within 24 h (<6 h) (<3 h) | A combination of caspase-3, D-dimer, sRAGE, chimerin, secretagogin, MMP-9 differentiated stroke from mimics. Best association with a model of caspase-3 and D-dimer | Validation phase not done | Caspase-3 52% D-dimer 81% sRAGE 48% Chimerin 79% Secretagogin 67% MMP-9 65% | Caspase-3 73% D-dimer 38% sRAGE 65% Chimerin 32% Secretagogin 48% MMP-9 65% |
4. | Dambinova (2012) (USA) [35] | 101—IS 71—No stroke 20—Mimics 52—HC 48—Risk factor controls | NR-2 peptide | Rapid magnetic particle ELISA | Within 72 h | NR-2 peptide distinguishes stroke from mimics | Validation phase not done, blood sampling within a narrow time window is required | 92% | 96% |
5. | Dassan (2012) (UK) [36] | 29—IS 15—Mimics 15—HC | Serum VEGF | ELISA | 24 h | VEGF has limited clinical utility for diagnosis of IS from mimics | Less sample size, validation phase not done | 69% | 73% |
6. | Gonzalez-Garcia (2012) (Cuba) [29] | 44—IS 11—Mimics 79—High-risk control | Serum NSE and S100B | Immunoassay | 12–48 h | NSE and S100B cannot efficiently distinguish stroke from mimics | Small sample size, validation phase not done | NSE 53% S100B 55% | NSE 64% S100B 64% |
7. | Knauer (2012) (Germany) [40] | 100—IS 49—Mimics | BNP, D-dimer, MMP-9, S100B | Sandwich Fluorescence Immunoassay Triage stroke panel | <6 h | No single or combination of biomarker distinguished IS from mimics | Low specificity of the biomarker panel | Cutoff: 2.3 86% Cutoff: 2.5 84% | Cutoff: 2.3 33% Cutoff: 2.5 35% |
8. | Wendt (2015) (Germany) [37] | 287—IS 90—Mimics | Copeptin | Immune fluorescent assay | 15 to more than 180 minutes | Copeptin does not discriminate between stroke and mimics | Sensitivity and specificity data not reported | – | – |
9. | Laskowitz (2009) (USA) [38] | 293—IS 361—Mimics Validated 87—IS 152—Mimics | MMP-9, D-dimer, S100B, BNP | Triage stroke panel (fluorescent immunoassay) | 9.3 h (4.5–18.2 h) | A model of MMP-9, D-dimer, S100B, BNP can differentiate stroke from mimics | The biomarker levels may be elevated due to presence of other comorbidities | First Quartile 86% third Quartile 35% | First Quartile 37% third Quartile 86% |
10. | Ahn (2011) (South Korea) [33] | 28—IS 24—Non-stroke/mimics | Albumin adjusted IMA index, IMA | Albumin Cobalt binding test | Within 6 h | IMA index more sensitive than conventional IMA to diagnose stroke | Small sample size, no validation phase | IMA Index 95.8% IMA 87.5% | IMA Index 96.4% IMA 89.3% |
11. | Airas (2008) (Finland) [31] | 20—IS 20—Mimics | sVAP-1 | Time resolved Immunofluorometric ELISA assay | Within 6 h | sVAP-1 levels higher in stroke patients as compared to age-sex matched control (mimic) group | Small sample size, no validation phase | – | – |
12. | Doehner (2012) (Germany) [32] | 124—IS 49—Non-stroke/mimics | PENK-A, PTA (plasma) | Sandwich ELISA | 4.5 h (1.3–19.8 h) | Only PENK-A levels were higher in stroke patients as compared to non-stroke group | Small sample size, validation in other populations required | – | – |
15. | Sibon (2009) (France) [39] | 126—Stroke 63—Mimics | Triage stroke panel MMX (D-dimer, BNP, MMP-9, S100β) | Fluorescence immunoassay | – | MMX stroke panel of four biomarkers distinguished stroke from mimics | Very low specificity, no individual differentiation of IS or HS from mimics | 92.9% (87–96.2%) | 23.8% (14.9–35.6%) |
16. | An (2013) (Korea) [30] | 188—Stroke 90—Mimics | NSE, VSNL-1, hFABP, Ngb, S100B, GFAP, IL-6, TNFα, MMP-9, PAI-1 | ELISA | 6–24 h | A panel of IL-6, S100B, MMP-9 distinguished stroke from mimics but with low discrimination ability | Weak discrimination ability of true stroke from mimics | – | – |
17. | Meng (2011) (China) [34] | 152—IS 46—Non-stroke/mimics | AT-III, TAT, Fibrinogen, D-dimer, hsCRP (Plasma) | Immunoturbidimetry assay, Chromogenic assay, ELISA | Within 4.5 h | Plasma AT-III and fibrinogen distinguished stroke from non-stroke | No validation phase performed, non-stroke group not clearly defined | AT-III 97.37% Fibrinogen 96.05% | AT-III 93.62% Fibrinogen 82.61% |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
