Buenos Aires 01 de Marzo del 2021
Blood biomarkers for the diagnosis and differentiation of stroke: A systematic review and meta-analysis
Blood biomarkers for the diagnosis and differentiation of stroke: A systematic review and meta-analysis
Shubham Misra, Joan Montaner, Laura Ramiro,Rohan Arora,Pumanshi
Talwar, Manabesh Nath,Amit Kumar, Pradeep Kumar, Awadh K Pand,
Dheeraj Mohania, Kameshwar Prasad and Deepti Vibha
International Journal of Stroke 2020, Vol. 15(7) 704–721
Since ‘‘time is brain,’’ rapid diagnosis of stroke and its subtypes is essential in early stages owing to the timesensitive nature of revascularization therapies. In recent years, advanced neuroimaging techniques have significantly improved the management and treatment of stroke. However, the diagnosis of stroke is often challenging in resource-limited settings where neuroimaging facilities and experienced neurologists are not readily available. A computed tomography (CT) scan has low sensitivity to detect an ischemic stroke (IS) and is often normal in the acute phase.1 Further, a delayed door-to- CT time has led to the underutilization of these timedependent therapies. Therefore, even today stroke diagnosis largely remains a clinical decisión
The plethora of research conducted in the last 20 years on determining a suitable blood-based protein biomarker for stroke diagnosis offers some hope toward an alternative approach. A potential blood biomarker for the diagnosis of IS should be able to differentiate an IS from intracerebral hemorrhage (ICH), stroke mimics, and transient ischemic attack (TIA).
Stroke mimics constitute a heterogeneous group of patients, which are difficult to identify and differentiate from IS using neuroimaging in the early hours. They have similar characteristic symptoms compared to stroke patients but have a nonvascular etiology. A biomarker should be able to differentiate IS from all similar symptomatic conditions to be clinically useful.
However, a biomarker need not be both highly sensitive and highly specific at the same time. As per the research question, a biomarker could either be used to rule out a condition (sensitivity) or to rule in a condition (specificity).
A blood biomarker should preferably be able to diagnose and differentiate IS within the 4.5-h window in order to maximize the benefits of the thrombolytic therapy. Recently, published trials such as DAWN3 and DEFUSE-34 have shown promising results in administering the endovascular thrombectomy in selected patients in the extended time window up to 24 h. Thus, blood biomarkers diagnosing IS within 24 h could also be of clinical importance in deciding stroke-type specific treatment.
Therefore, our systematic review and meta-analysis aims to pool the evidence generated to date for determining blood-based protein biomarkers in diagnosing and differentiating IS from healthy controls, stroke mimics, TIA, and ICH within 24 h of symptom onset
Dado que ''el tiempo es cerebro'', el diagnóstico rápido de accidente cerebrovascular y sus subtipos es esencial en las primeras etapas debido a la naturaleza sensible al tiempo de las terapias de revascularización.
En los últimos años, las técnicas avanzadas de neuroimagen han mejorado significativamente el manejo y tratamiento del accidente cerebrovascular. Sin embargo, el diagnóstico de accidente cerebrovascular a menudo es difícil en entornos con recursos limitados donde las instalaciones de neuroimagen y los neurólogos experimentados no están fácilmente disponibles. Una tomografía computarizada (TC) tiene baja sensibilidad para detectar un accidente cerebrovascular isquémico (IS) y a menudo es normal en la fase aguda.1 Además, un tiempo de puerta a TC retrasado ha llevado a la infrautilización de estas terapias dependientes del tiempo. Por lo tanto, incluso hoy en día el diagnóstico de accidente cerebrovascular sigue siendo en gran medida una decisión clínica
La plétora de investigaciones llevadas a cabo en los últimos 20 años sobre la determinación de un biomarcador proteico adecuado a base de sangre para el diagnóstico de accidente cerebrovascular ofrece cierta esperanza hacia un enfoque alternativo. Un biomarcador sanguíneo potencial para el diagnóstico de EI debe ser capaz de diferenciar un IS de hemorragia intracerebral (ICH), imitaciones de accidente cerebrovascular y ataque isquémico transitorio (AIT).
Las imitaciones de accidentes cerebrovasculares constituyen un grupo heterogéneo de pacientes, que son difíciles de identificar y diferenciar de EI mediante neuroimagen en las primeras horas. Tienen síntomas característicos similares en comparación con los pacientes con accidente cerebrovascular, pero tienen una etiología no vascular. Un biomarcador debe ser capaz de diferenciar el IS de todas las condiciones sintomáticas similares para ser clínicamente útil.
Sin embargo, un biomarcador no tiene por qué ser altamente sensible y muy específico al mismo tiempo. Según la pregunta de investigación, un biomarcador podría utilizarse para descartar una condición (sensibilidad) o para gobernar en una condición (especificidad).
Un biomarcador sanguíneo debe ser preferiblemente capaz de diagnosticar y diferenciar IS dentro de la ventana de 4.5-h con el fin de maximizar los beneficios de la terapia trombolítica. Recientemente, ensayos publicados como DAWN3 y DEFUSE-34 han mostrado resultados prometedores en la administración de la trombectomía endovascular en pacientes seleccionados en el período de tiempo extendido hasta 24 h. Por lo tanto, los biomarcadores sanguíneos que diagnostican IS dentro de 24 h también podrían ser de importancia clínica para decidir el tratamiento específico de tipo cerebrovascular.
Por lo tanto, nuestra revisión sistemática y metanálisis tiene como objetivo agrupar la evidencia generada hasta la fecha para determinar los biomarcadores proteicos a base de sangre en el diagnóstico y diferenciación de EI de controles saludables, imitaciones de accidente cerebrovascular, TIA, y ICH dentro de 24 horas de inicio de síntomas
Correct diagnosis of stroke and its subtypes is pivotal in early stages for optimum treatment.
The aim of this systematic review and meta-analysis is to summarize the published evidence on the potential of blood biomarkers in the diagnosis and differentiation of stroke subtypes.
A literature search was conducted up to 20 April 2020 in PubMed, EMBASE, Medline, Cochrane Library,
TRIP Database, and Google Scholar databases to search for eligible studies detailing the role of blood
biomarkers in diagnosing stroke. The following MeSH or free text terms were used to search the databases: (‘‘blood biomarkers’’ OR ‘‘protein biomarkers’’) AND (‘‘ischemic stroke’’ OR ‘‘IS’’) AND (‘‘intracerebral hemorrhage’’ OR ‘‘‘ICH’’) AND (‘‘transient ischemic attack’’ OR ‘‘TIA’’) AND (‘‘stroke mimics’’) AND (‘‘healthy controls’’) AND (‘‘diagnosis’’ OR ‘‘differentiation’’).
For the detailed search strategy refer to Suppl. File
1. The search was restricted to studies performed on human subjects. No restriction on language or publication period was set. The reference list from all included studies was also searched thoroughly for any additional eligible article. The protocol for this systematic review and meta-analysis was registered in PROSPERO (ID: CRD42019139659) and there were no major deviations from the published protocol in PROSPERO.
Individuals with acute stroke symptoms and controls aged >18 years.
Blood-based protein biomarkers obtained within 24 h of symptom onset to differentiate IS from various
Blood biomarker levels in IS patients were compared to blood biomarker levels in healthy controls, stroke mimics, TIA, and ICH.
Association of blood-based protein biomarker levels between IS and healthy controls, stroke mimics, TIA, and ICH within 24 h of symptom onset.
(i) Association of blood-based protein biomarker levels between total stroke (IS and ICH) and
healthy controls, stroke mimics, and TIA.
(ii) Association of blood-based protein biomarker levels between IS and healthy controls, stroke mimics,
TIA, and ICH within 6 h of symptom onset.
(iii) Association of blood-based protein biomarker levels between TIA and healthy controls and stroke mimics within 24 h of symptom onset.
A literature search was conducted for papers published until 20 April 2020 in PubMed, EMBASE, Cochrane Library, TRIP, and Google Scholar databases to search for eligible studies investigating the role of blood biomarkers in diagnosing stroke. Quality assessment was done using modified Quality Assessment of Diagnostic Accuracy Studies questionnaire. Pooled standardized mean difference and 95% confidence intervals were calculated. Presence of heterogeneity among the included studies was investigated using the Cochran’s Q statistic and I2 metric tests. If I2 was<50% then a fixed-effect model was applied else a random-effect model was applied. Risk of bias was assessed using funnel plots and between-study heterogeneity was assessed using meta-regression and sensitivity analyses. Entire statistical analysis was conducted in STATA version 13.0.
The initial search yielded 712 articles by searching various databases. After screening 622 records, 85 full text articles were assessed for eligibility out of which a total of 40 studies were included in our systematic review and meta-analysis consisting of 5001 IS,756 ICH, 554 stroke mimics, and 1774 healthy control subjects.Eight studies were included from Germany,11–18 seven were included from Spain,19–25 four studies each from China,26–29 Republic of Korea30–33 and Italy,34–37 two studies from India38,39 and one study each from Greece,40 Turkey,41 Czech Republic,42 United States of America,43 Taiwan,44 Sweden,45 Israel,46 Kuwait,47 Croatia,48 Norway,49 and Japan50 were included in our meta-analysis. The publication year of the included studies ranged from 1991 to 2020.
A total of 40 studies including patients with 5001 ischemic strokes, 756 intracerebral hemorrhage, 554 stroke mimics, and 1774 healthy control subjects analyzing 25 biomarkers (within 24 h after symptoms onset/after the event) were included in our meta-analysis; 67.5% of studies had moderate evidence of quality. Brain natriuretic peptide, matrix metalloproteinase-9, and D-dimer significantly differentiated ischemic stroke from intracerebral hemorrhage, stroke mimics, and health control subjects (p<0.05). Glial fibrillary acidic protein successfully differentiated ischemic stroke from intracerebral hemorrhage (standardized mean difference _1.04; 95% confidence interval _1.46 to _0.63) within 6 h. No studies were found to conduct a meta-analysis of blood biomarkers differentiating transient ischemic attack from healthy controls and stroke mimics.
Our systematic review and meta-analysis assessed the potential of 25 blood biomarkers in 40 studies for the diagnosis of IS within 24 h and observed three biomarkers BNP, MMP-9, and D-dimer that significantly differentiated IS from healthy controls, stroke mimics, and ICH. GFAP successfully differentiated IS from healthy controls and ICH but not from stroke mimics. S100B, Caspase-3, and NSE could only differentiate IS from stroke mimics, UCH-L1 only differentiated IS from ICH while IL-6, CRP, hFABP, and TNF-a differentiated IS from healthy controls only.
To the best of our knowledge, this is the first systematic review and meta-analysis which additionally
pooled the available evidence for (i) biomarkers differentiating total stroke (ischemic þ hemorrhagic) from healthy controls and stroke mimics as well as (ii) for the biomarkers analyzed within 6 h for the diagnosis and differentiation of IS. Out of the six biomarkers identified, D-dimer and GFAP significantly
differentiated total stroke from healthy controls and stroke mimics. Further, analysis conducted within six
hours reiterated the potential of GFAP in successfully differentiating IS from healthy controls and ICH.
Within 6 h, D-dimer differentiated IS from stroke mimics whereas BNP differentiated IS from ICH,
In the last 10–15 years, numerous systematic reviews have been published and a huge evidence has been generated regarding the role of diagnostic blood biomarkers in stroke.6,53–55 GFAP, the most extensively studied protein, has been identified as a potential diagnostic biomarker in several systematic reviews and meta-analyses for successfully differentiating ICH from IS.56–59 The results of our meta-analysis are also in concordance with the previously published metaanalyses on GFAP. A systematic review published by Monbailliu et al. in 2017 identified BNP and S100B as potential biomarkers for the diagnosis of IS.54 Contrary to the studies published thus far, they did not find GFAP as a potential biomarker to differentiate IS from ICH. The results of our meta-analysis support the results of Monbailliu et al. in terms of BNP biomarker but differ significantly in terms of S100B and GFAP. This difference in findings could mainly be due to the smaller number of studies included in the Monbailliu et al. review. They included only four studies for examining the diagnostic potential of GFAP biomarker in differentiating IS from ICH whereas in our analysis we included 10 studies. Therefore, their study might be underpowered to observe the desired effect. Similarly, as compared to the Monbailliu et al.
review, we did not find S100B as a promising biomarker in differentiating IS from healthy controls and ICH within 24 hs. Since the number of studies pooled for S100B biomarker were more in our study (two studies for healthy controls and seven for ICH) in contrast to
the review published by Monbailliu et al. (one study for healthy controls and two for ICH), the significant relationship might have been lost when more studies were pooled.
Despite the large number of reviews published, certain inherited limitations present in the methodology
and/or systematic conduct of these reviews have decreased the quality of evidence generated. To overcome these limitations in our study, we carefully explored the heterogeneity and biases present in each article. We used SMD as the summary statistics to pool the results instead of mean difference (MD) used in the previous reviews.53,54 SMD has more external applicability and generalizability and thus is a preferred summary measure over MD.60 We tried to minimize the selection bias by excluding the studies which mixed IS cases with TIA61 and/or TIA with mimics/non-stroke62 so as to avoid any unfair comparisons. In contrast to the reviews published earlier,53,54 we avoided pooling the levels of different biomarkers together, for instance CRP and hsCRP. hsCRP is a highly sensitive biomarker and must be analyzed separately from CRP to reduce the heterogeneity and increase the specificity of the findings. As compared to the systematic review by Monbailliu et al.,54 we did not include a ‘‘patiens control’’ reference group in our analysis owing to the heterogeneous population present in the group.
Even though we took extensive measures to avoid any bias and reduce the heterogeneity at every step, it would be unfair to interpret our findings without considering the following limitations. The mean and SD levels of biomarkers in few studies were converted from the actual reported median and IQR levels, therefore they did not represent the exact mean and SD values of the biomarkers. Considerable amount of heterogeneity was present in some of the biomarkers analyzed in our meta-analysis. However, we conducted meta-regression and sensitivity analysis to explain for the source of heterogeneity and used a random-effect model wherever the heterogeneity was high. Since some biomarkers have been extensively studied than the others, the presence of publication bias cannot be ruled out completely.
This meta-analysis highlights the potential of BNP, MMP-9, D-dimer, and GFAP as diagnostic biomarkers
for stroke within 24 h.
Our meta-analysis highlights 11 biomarker candidates that might have the potential to differentiate IS from various conditions within 24 h. It also generates the evidence for potential biomarker candidates to differentiate total stroke from controls and mimics within 24 h as well as the evidence to differentiate IS within 6 h of symptom onset. The results of our meta-analysis might serve as a platform for conducting further targeted proteomics studies and phase-III clinical trials.