Archives of Academic Emergency Medicine. 2023; 11(1): e9 REV I EW ART I C L E Serum Glial Fibrillary Acidic Protein in Detecting In- tracranial Injuries Following Minor Head Trauma; a Sys- tematic Review and Meta-Analysis Sajjad Ahmadi1, Shayan Roshdi Dizaji2, Asrin Babahajian3, Mohammadreza Alizadeh2, Arash Sarveazad4,5∗, Mahmoud Yousefifard2,6 † 1. Emergency Medicine Research Team, Faculty of Medicine, Tabriz University of Medical Sciences, Tabriz, Iran. 2. Physiology Research Center, Iran University of Medical Sciences, Tehran, Iran. 3. Liver and Digestive Research Center, Research Institute for Health Development, Kurdistan University of Medical Sciences, Sanandaj, Iran. 4. Colorectal Research Center, Iran University of Medical Sciences, Tehran, Iran. 5. Nursing Care Research Center, Iran University of Medical Sciences, Tehran, Iran. 6. Pediatric Chronic Kidney Disease Research Center, Tehran University of Medical Sciences, Tehran, Iran. Received: October 2022; Accepted: December 2022; Published online: 1 January 2023 Abstract: Introduction: Developing novel diagnostic and screening tools for exploring intracranial injuries following mi- nor head trauma is a necessity. This study aimed to evaluate the diagnostic value of serum glial fibrillary acidic protein (GFAP) in detecting intracranial injuries following minor head trauma. Methods: An extensive search was performed in Medline, Embase, Scopus, and Web of Science databases up to the end of April 2022. Hu- man observational studies were chosen, regardless of sex and ethnicity of their participants. Pediatrics studies, report of diagnostic value of GFAP combined with other biomarkers (without reporting the GFAP alone), arti- cles including patients with all trauma severity, defining minor head trauma without intracranial lesions as the outcome of the study, not reporting sensitivity/specificity or any other values essential for computation of true positive, true negative, false positive and false-negative, being performed in the prehospital setting, assessing the prognostic value of GFAP, duplicated reports, preclinical studies, retracted articles, and review papers were excluded. The result was provided as pooled sensitivity, specificity, diagnostic score and diagnostic odds ratio, and area under the summary receiver operating characteristic (SROC) curve with a 95% confidence interval (95% CI). Results: Eventually, 11 related articles were introduced into the meta-analysis. The pooled analysis implies that the area under the SROC curve for serum GFAP level in minor traumatic brain injuries (TBI) was 0.75 (95% CI: 0.71 to 0.78). Sensitivity and specificity of this biomarker in below 100 pg/ml cut-off were 0.83 (95% CI: 0.78 to 0.89) and 0.39 (95% CI: 0.24 to 0.53), respectively. The diagnostic score and diagnostic odds ratio of GFAP in detection of minor TBI were 1.13 (95% CI: 0.53 to 1.74) and 3.11 (95% CI: 1.69 to 5.72), respectively. The level of evidence for the presented results were moderate. Conclusion: The present study’s findings demonstrate that serum GFAP can detect intracranial lesions in mild TBI patients. The optimum cut-off of GFAP in detection of TBI was below 100 pg/ml. As a result, implementing serum GFAP may be beneficial in mild TBI diagnosis for preventing unnecessary computed tomography (CT) scans and their related side effects. Keywords: Brain Injuries, Traumatic; Diagnosis; Biomarkers Cite this article as: Ahmadi S, Roshdi Dizaji S, Babahajian A, Alizadeh M, Sarveazad A, Yousefifard M. Serum Glial Fibrillary Acidic Protein in Detecting Intracranial Injuries Following Minor Head Trauma; a Systematic Review and Meta-Analysis. Arch Acad Emerg Med. 2023; 11(1): e9. https://doi.org/10.22037/aaem.v11i1.1682. ∗Corresponding Author: Arash Sarveazad; Colorectal Research Center, Rasoul-e-Akram Hospital, Nyaiesh Ave., Tehran, Iran; Tel/Fax: +982166554790; Email: Arashsarveazad@gmail.com, ORCID: https://orcid.org/0000-0001- 9273-1940. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index S. Ahmadi et al. 2 1. Introduction Traumatic brain injury (TBI) is the leading cause of death and debilitation among the young population worldwide (1). Statistics demonstrate its worldwide prevalence at 8.4%, which has been considered a major health concern in devel- oping countries. The incidence rate and disability-adjusted life years (DALY) of TBI increased significantly from 1990 to 2016. TBI is a significant burden on the healthcare sys- tem and constitutes a large portion of hospital visits and ad- missions (2). Based on clinical assessments, and typically, the Glasgow coma scale (GCS), TBI patients are commonly classified into three subgroups: mild (GCS:13-15), moderate (GCS:9-12), and severe (GCS:8-12), with mild TBI accounting for more than 85% of cases (3, 4). The intracranial lesions caused by TBI are conventionally de- tected using imaging techniques, in particular, non-contrast- enhanced computed tomography (CT) scan and rarely, mag- netic resonance imaging (MRI) (5, 6). However, the role of imaging techniques has been debated due to negative find- ings in a vast majority of mTBI patients and the consider- able amount of hazardous ionizing radiation associated with it and cost implications imposed by it (7). According to avail- able data, only 16% of mild TBI patients have demonstrable intracranial lesions on computed tomography (CT) imaging (8). To prevent unnecessary imaging, multiple clinical deci- sion rules with varied diagnostic values have been developed to identify those who are at risk of having pathoanatom- ical intracolonial lesions (9). One limitation with current decision rules is the presence of clinical criteria, including headache, post-traumatic amnesia, nausea, vertigo, dizzi- ness, and consciousness assessment that are self-reporting or non-specific and subject to clinical examiner bias. The demand for minimally invasive objective parameters in detecting brain injury has led to increasing attention toward serum biomarkers serving as supplementary screening tools. In light of advancements in understanding the molecular bi- ology and pathways involved in neuronal damage, S100-B and neuron specific-enolase (NSE), two brain-enriched cy- toplasmic markers, were widely studied to aid in diagnosing central nervous system injuries (1, 10, 11). Although demon- strated to pose promising efficacy in providing insight into injury severity and progression, these biomarkers are sub- ject to some drawbacks. Studies on multiple trauma have re- vealed that extracranial sources of NSE and S100-B release confound its interpretation and impede its practical clinical † Corresponding Author: Mahmoud Yousefifard; Physiology Research Center, Iran University of Medical Sciences, Hemmat Highway, P.O Box: 14665-354, Tehran, Iran; Phone/Fax: +982186704771; Email: yousefifard.m@iums.ac.ir, ORCID: https://orcid.org/0000-0001-5181-4985. utility in acute settings (12, 13). Among emerging serum biomarkers, studies outlined glial fibrillary acidic protein (GFAP) as a brain-specific astroglia cytoskeleton filament that is released into peripheral circu- lation upon neuronal injuries due to patency of the blood- brain barrier (BBB) (14). Clinical investigations have demonstrated that the GFAP serum level has a high correlation with the lesion volume and severity in traumatic brain injuries. This made GFAP a poten- tial candidate for diagnostic and prognostic purposes in TBI evaluation. Although the evidence has supported that GFAP holds the potential to predict intracranial injuries in TBI patients, there is yet considerable heterogeneity among the available studies (9, 15, 16). Given the inconsistencies found between stud- ies and the lack of a consensus, this systematic review with meta-analysis has been carried out to draw a conclusion on the diagnostic utility of GFAP in detecting intracranial lesions following mild TBI. 2. Methods 2.1. Study design The present study is a systematic review to investigate the diagnostic value of GFAP in detecting traumatic intracranial lesions following TBI. An extensive search was performed in available electronic resources and databases to achieve this aim and discover all the relevant papers. The current study design adhered to MOOSE guidelines (17). The PICO descrip- tion of the present study is shown below: The Problem or Population (P) consists of human studies performed on patients with mild TBI. Mild TBI was defined as a GCS between 13 to 15. The index test (I) was the venous serum value of GFAP. The comparison (C) was done with ref- erence standards such as CT scans, MRI, and standard guide- lines for detecting sport-related concussions. The outcome (O) was intracranial lesions. We have not previously registered our study protocol in any registries. 2.2. Search strategy The keywords were selected by reviewing the MeSH and Emtree terms for Medline (via PubMed) and Embase databases, respectively. Other possibly related keywords were found by consulting the experts in the field and scru- tinizing the jargon and related article titles and keywords. Subsequently, using the appropriate combination of the key- words, the exploration was executed in Medline, Embase, Scopus, and Web of Science databases from their inception to the end of April 2022. The detailed search queries on each database are presented in appendix 1. In addition, a man- ual search was done on the bibliography of related articles This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index 3 Archives of Academic Emergency Medicine. 2023; 11(1): e9 and results obtained from Google and Google Scholar motor search engines. 2.3. Eligibility criteria Diagnostic accuracy studies performed on the serum value of GFAP in mild TBI patients were considered relevant. Hu- man studies on adult patients (age>15 years old) were cho- sen, regardless of participants’ sex and ethnicity. Exclusion criteria included pediatrics studies, report of diagnostic value of GFAP combined with other biomarkers (without reporting the GFAP alone), no discrimination of TBI severity based on GCS, mild TBI without intracranial lesions as the outcome of the study, being performed in the prehospital setting, as- sessing the prognostic value of GFAP, and preclinical studies. We also excluded duplicated reports, reports with inadequate data on sensitivity/specificity or any corresponding values essential for computation of true positive (TP), true negative (TN), false positive (FP), and false-negative (FN), retracted articles, and reviews. We applied no restrictions on the lan- guage, geographical location, or publication year during our survey. We also excluded studies assessing the biomarker of interest qualitatively. If the serial serum level of GFAP was evaluated in a study, the first observation was included in the meta-analysis. 2.4. Screening and data extraction Records collected during systematic and manual searches were exported to Endnote version 20.0 software (Clarivate Analytics, Philadelphia, PA, USA), and duplicates were re- moved. Two independent researchers screened the records by their titles and abstracts and determined those deemed as possibly related and retrieved the full text of relevant articles. Any discordances were resolved by discussion or consulting the third researcher. Eligible studies based on the above- mentioned inclusion and exclusion criteria were included for data collection. We extracted the data on study information (design, first author name, study time and location, publication year), patients’ characteristics (sample size and demographics, level of consciousness based on GCS), outcome definition and standard reference, overall number and demographics of outcome and non-outcome groups, serum GFAP assay method, the time elapsed from the traumatic accident un- til blood acquisition, recommended GFAP cut-off points and associated sensitivity and specificity. Whenever two or more articles were extracted from the same registry, the article with larger sample size or longer follow-up duration was selected to be included. Additionally, if evaluated values were re- ported in distinct subgroups (e.g., age groups, sex groups, etc.), the data was documented and analyzed corresponding to that group. Reviewers manually entered all the extracted data into a predefined checklist in an Excel sheet (Microsoft, Redmond, Washington, USA). 2.5. Risk of bias assessment The methodological quality of included articles was assessed using the Quality Assessment of Diagnostic Accuracy Stud- ies Version 2 (QUADAS-2) tool (18). This implementation in- volves risk of bias and applicability, which were evaluated and reported in each article by two independent researchers. QUADAS-2 instruction classifies scores into three levels: low risk, high risk, and unclear. Any disagreements were ad- dressed by discussion or consulting with the third researcher. 2.6. Level of evidence Two independent reviewers determined the level of evidence for our primary interest outcome using the Grading of Rec- ommendations, Assessment, Development, and Evaluations (GRADE) framework (19). The GRADE framework consists of subjective evaluation of five domains of risk of bias, impreci- sion, inconsistency, indirectness, and publication bias. 2.7. Statistical analysis Analyses were performed using STATA 17.0 statistical soft- ware. The studies were summarized and sorted based on their diagnostic value. Attaining the TP, TN, FP, and FN values was the priority in the data collection process. If the afore- mentioned data was not reported in the articles, the authors were contacted through email and asked to provide the re- quired data. Finally, in the case of the authors’ unresponsive- ness, the data were extracted by employing analytics on re- ported sensitivity and specificity. Analyses were performed using the "Midas" package of STATA, which pools article di- agnostic value data using a bivariate mixed-effects binary re- gression modeling framework. The result was provided as pooled sensitivity, specificity, diagnostic score and diagnos- tic odds ratio, and area under the summary receiver operat- ing characteristic curve (SROC) with a 95% confidence inter- val (95% CI). To measure the heterogeneity between studies, I2 statis- tics were performed. Whenever heterogeneity (I2>50%) was found, subgroup analysis was carried out to investigate the source of heterogeneity. Since three studies included a small portion of moderate TBI patients (1.0% to 3.6% of their to- tal sample size), a sensitivity analysis was applied to stud- ies that included only mild TBI patients. In addition, a sen- sitivity analysis was performed according to the definition of the non-TBI group and the setting of patients (mild-TBI vs. sport-related concussion). Meta-analysis was performed when the required data were reported in at least four distinct analyses. Publication bias was assessed using Deek’s asym- metry funnel plot test. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index S. Ahmadi et al. 4 3. Results 3.1. Screening and specificity of included papers The search in databases led to 4475 articles. After removing the duplicate items, abstracts of 2352 papers were screened. In the next step, full texts of 84 articles were obtained for a more thorough evaluation. Eventually, 11 related articles (20- 30) were introduced into the meta-analysis. Figure 1 repre- sents the inclusion process and reasons for exclusions. The eventual eligible papers included 8 cohort studies, 2 clinical trials, and 1 cross-sectional study. All studies were conducted on suspected or confirmed TBI patients who at- tended the hospital. The control group was composed of CT-negative patients in 10 papers, and both CT-negative and healthy individuals in 1 study. These studies included 4978 individuals (907 TBI patients and 4071 non-TBI subjects). The average age of included patients was 24 to 56.0 years old. Diagnosis of TBI was achieved employing CT scan or MRI in all studies. In all studies, the time interval between head trauma and GFAP measurement was less than 24 hours post-injury. Eight studies investigated mild TBI (21-25, 27, 28, 30), and 3 ex- plored mild to moderate injuries (20, 26, 29). It is worth men- tioning that moderate TBI cases comprised a small portion of all participants (1-3.6% of the whole population), so the study was mainly performed on mild TBI cases. It must be noted that in the included studies, data were reported in var- ious subgroups and different GFAP cut-off points; therefore, data were formulated into 30 distinctive analyses in the cur- rent meta-analysis. Table 1 depicts a summary of included studies’ characteristics. 3.2. Risk of bias To investigate the risk of bias, the QUADAS-2 instrument was utilized. The risk of bias in all articles was categorized as low risk (Table 2). In conclusion, the overall risk of bias score was low. 3.3. Diagnostic value of GFAP in detecting in- tracranial lesions in mild TBI The pooled analysis implies that the area under the SROC curve for GFAP serum level in mild TBI is 0.75 (95% CI: 0.71 to 0.78) (Figure 2). Sensitivity and specificity of this biomarker were calculated as 0.83 (95% CI: 0.78 to 0.87) and 0.38 (95% CI: 0.27 to 0.59), respectively (Figure 3). Diagnostic score and diagnostic odds ratio of GFAP in the detection of intracranial lesions in mild TBI were calculated to be 1.13 (95% CI: 0.53 to 1.74) and 3.11 (95% CI: 1.69 to 5.72), respectively (Figure 4). 3.4. Subgroup analysis Subgroup analyses were performed in response to apparent heterogeneity among articles in analyses (I2 value between 91.55% and 100%). The findings of this analysis are presented in table 3. As was speculated, the type of the study had a profound impact on the findings. After performing sepa- rate analyses based on study type, the heterogeneity between studies decreased significantly and was figured to be 0.0%. Over and above that, the value for GFAP sensitivity in detec- tion of intracranial lesions in mild TBI was evaluated as 0.62 (95% CI: 0.25, 1.00) for cross-sectional studies and 0.84 (95% CI: 0.79, 0.89) for cohorts, which can be concluded that cross- sectional studies had the least reported value among other studies. Another subgroup analysis was performed based on reported GFAP cut-off points. The reported cut-off points varied from 3.0 to 848 pg/mL. As a result, they were grouped into three sets with values of less than 100 pg/mL, 100-300 pg/ml, and above 400 pg/mL. Analyses revealed that another source of heterogeneity among studies was variability in cut-off points recommended by studies. Grouping of analyses based on cut-off points led to a reduction in the heterogeneity to 0%. The sensitivity and specificity of GFAP in cut-offs below 100 pg/mL were estimated to 0.83 (95% CI: 0.78 to 0.89) and 0.39 (95% CI: 0.24 to 0.53), respectively. Additionally, the sensi- tivity and specificity of this biomarker in cut-offs between 100-300 pg/mL were 0.79 (95% CI: 0.67 to 0.90) and 0.32 (95% CI: 0.08 to 0.57), respectively. Finally, in cut-offs above 400 pg/mL the sensitivity and specificity were measured as 0.89 (95% CI: 0.81 to 0.96) and 0.43 (95% CI: 0.15 to 0.71), respectively. Seemingly, in cut-offs below 100 pg/mL and above 400 pg/mL, GFAP has the optimum performance in detecting in- tracranial lesions among mild TBI patients. There were 17 vs. 5 analyses in the group with cut-offs below 100 pg/mL and above 400 pg/mL, respectively. As a result, a cut-off below 100 pg/mL seems to be associated with more reliable diag- nostic accuracy for GFAP in identifying intracranial lesions in mild TBI cases (Table 3). 3.5. Sensitivity analysis In addition to subgroup analysis, sensitivity analysis was per- formed in the current study. Sensitivity analysis on studies in which the control group consisted of CT-negative patients re- vealed that the difference in the control group also does not influence sensitivity (sensitivity=0.84; p=0.16) and specificity (specificity=0.40; p=0.86) of GFAP in diagnosis of intracranial lesions of mild TBI patients. On the contrary, sensitivity anal- ysis indicated that the studies that exclusively recruited mild TBI cases have reported significantly lower sensitivity than studies on mild to moderate TBI patient populations (0.80 vs. 0.95; p<0.001). However, it has no effect on specificity of GFAP (0.38 vs. 0.42; p=0.93) (Table 4). This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index 5 Archives of Academic Emergency Medicine. 2023; 11(1): e9 3.6. Publication bias Deek’s asymmetry funnel plot test was performed to assess publication bias. The analysis implied no possible publica- tion bias in the included studies (p=0.97) (Figure 5). 3.7. Certainty of evidence There was high heterogeneity among included study, but our subgroup analysis could find the sources of heterogeneity. The overall grade was rated up one point due to large mag- nitude of effect size (based on diagnostic odds ratio). So, the level of evidence was graded moderate (Table 4). 4. Discussion Our findings demonstrated that GFAP serum level is capa- ble of detecting intracranial lesions in mild TBI patients (AUC:0.75 with 95% CI: 0.71 to 0.78). Our pooled analy- sis indicated that GFAP has a promising diagnostic value with 83% sensitivity and 39% specificity in cut-offs below 100 pg/ml. Utilizing serum GFAP measurement may aid health- care providers in being more selective in imaging acquisition, while providing safe and efficient care in the emergency de- partment and acute care settings. Measuring serum GFAP as a traumatic brain injury signature would lower the rate of un- necessary ionizing imaging and shorten the time spent for in-patient observation in the high-burden ED. Based on our investigations, the proposed cut points for GFAP among studies varied in a range of 3-848 pg/ml. As a result, we categorized these cut points into 3 groups with values of less than 100 pg/ml, 100-300 pg/ml, and above 400 pg/ml based on the fact that the increase in cut point results in lower sensitivity and higher specificity (31). The analyses indicated that the serum level of GFAP has the optimum per- formance in cut-offs below 100 pg/ml and above 400 pg/ml. However, only 5 studies reported cut-offs above 400 pg/ml, compared to 17 studies on cut-offs below 100 pg/ml. There- fore, we can only rely on findings of cut-offs below 100 pg/ml, and there is a need to investigate the optimum cut-offs in fu- ture studies. Application of imaging for diagnosis of sport-related concus- sion is limited when considering the fact that athletes sustain multiple head traumas during their sport career. So, diagno- sis of this type of mild TBI is based on clinical judgment of team physicians and recommended guidelines. Meier et al. reported that GFAP is not significantly elevated from baseline preinjury level following sport-related concussion in football players, except for athletes with loss of consciousness and amnesia (32). Asken et al. concordantly showed moderate diagnostic accuracy of GFAP (AUC:0.67, 95%CI:0.57–0.78) for sport-related concussion and demonstrated superiority of S100b (AUC:0.72, 95%CU:0.63–0.81) in this regard (33). It can be concluded that this biomarker holds moderate potential in detecting clinically diagnosed sports-related concussions. Nevertheless, addition of objective brain biomarkers would be beneficial in diagnosis of concussion among athletes who have minimal or less severe self-reporting symptoms and thus, imposing a challenge for physicians. Objective assess- ment of biomarkers can reduce the reliance on symptoms for diagnosis of concussion as some athletes may deliberately not report them to avoid being prohibited from participa- tion in the competitions (32). Also, the rise of biomarkers fol- lowing concussion can represent the pathophysiological re- sponses underlying the concussion and could provide prog- nostic information regarding the required time before clear- ance for full return to sport even when symptoms are allevi- ated. However, only 2 articles on the diagnostic accuracy of GFAP in sport-related concussion were found in our search, and further investigations are suggested in this regard. The reference standard imaging in many studies was the CT scan, and this would affect the judgment of the sensitiv- ity and specificity of GFAP, as some forms of brain damage, such as micro-injuries and diffuse axonal injury (DAI), are not conspicuously visible in CT scans and require more ad- vanced neuroimaging modalities such as MRI (34). More- over, in the current meta-analysis, there were 3 studies on mild to moderate TBI patients. However, there were 1%, 1.6%, and 3.6% of patients in the moderate TBI group. Seem- ingly, a significant proportion of patients were in the mild TBI group; hence, including these studies in the pooled anal- ysis is justifiable. We believe that akin to cardiac troponin for diagnosis of myocardial damage, a singular value of initial post-traumatic GFAP may not be able to provide concise and valid information on the extent of the patient’s central ner- vous system (CNS) damage. Serial measurements of GFAP could be more informative since its concentration pattern, and bulk release can shed light on the dynamics of brain in- jury evolution and identification of delayed injuries. Addi- tionally, the mere serum value of GFAP is not capable of char- acterizing different types of brain lesions (i.e., skull fractures, subdural hematoma, contusion, sub-arachnoid hemorrhage, etc.) and development of a model consisting of a panel of blood biomarkers originating from multiple types of cells en- gaged in TBI is under investigation. In 2018, the Food and Drug Administration (FDA) approved the clinical application of a semiquantitative kit developed by Banyan Biomarkers, Inc. for the measurement of GFAP and UCH-L1 to reduce the overuse of CT imaging in TBI (13). The concurrent measure- ment of UCH-L1 and GFAP enhanced their separate diagnos- tic values and 97.6% sensitivity and 37.4% specificity were re- ported when a combination assessment was performed (20). Other studies evaluated the incorporation of inflammatory, endothelial, and neural damage markers to reach a more fa- vorable diagnostic accuracy. However, this multiplex assess- ment is associated with higher costs compared to the CT scan This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index S. Ahmadi et al. 6 and constrains health resources (24, 35). The value of inte- grating blood biomarkers into the existing clinical decision rules also deserves further research. Kinetically, GFAP rises immediately after TBI with a peak level reached during the first 24 hours following injury and declines to near normal value in the next 14 to 30 days post- injury (22, 36). Although GFAP persists in circulation for a long time, the majority of studies included in this review had a time interval of less than 24 hours between traumatic acci- dent and blood draw; hence, it is speculated that the delayed measurement of GFAP may yield different diagnostic value (4, 37). Moreover, the primary laboratory technique for GFAP assay was reported to be ELlSA, which could take time for re- sults to be prepared. Further studies are required to endorse the clinical utility of GFAP at bedside evaluation by introduc- ing easy and rapid measurement techniques. It should be noted that the presented cut-offs and associ- ated sensitivity and specificity cannot be adopted for pedi- atric TBI or birth-inflicted brain injuries, since we primarily focused on adults. Additionally, caution should be imple- mented on the approach to GFAP values in patients with neu- ropsychological or renal function impairment, as these two disease categories may impact the serum GFAP concertation as a source of production and excretion, respectively. This brought up concerns about the reduced accuracy of GFAP in the diagnosis of intracranial injuries among the geriatric population, who suffer a higher rate of comorbidities and, are significantly more prone to harbor traumatic lesions yet symptomatically silent on presentation. To overcome this challenge, Gardner et al. recommended the integration of Tau level measurement to improve the diagnostic accuracy of GFAP in elder patients (38). Finally, we didn’t aim to ex- plore the cost-effectiveness of the proposed biomarker as- sessment, but it is worth stating that some studies reported controversial results on the costs of TBI biomarkers, and fur- ther studies in different contexts and healthcare systems are warranted to reach a common conclusion (38, 39). 5. Conclusions The findings of the current review demonstrated that with a moderate level of evidence, GFAP serum level is capable of detecting intracranial lesions in mild TBI patients and might facilitate the clinical decision-making and delivery of tar- geted therapeutic care. Additionally, sensitivity and speci- ficity of GFAP in cut-offs below 100 pg/ml were measured to be 0.83 and 0.42, respectively. Taken together, the measure- ment of GFAP after injury may be beneficial in mild TBI di- agnosis and prevents unnecessary CT scans if used as a com- plementary indicator to individual patient clinical character- istics and examinations. 6. Declarations 6.1. Acknowledgments Not applicable. 6.2. Ethics approval Not applicable. 6.3. Patient consent Not applicable 6.4. Informed consent Not applicable. 6.5. Availability of data and materials All data generated or analyzed during this study are included in this published article and its supplementary information file. 6.6. Permission to reproduce material from other sources Not applicable. 6.7. Clinical trial registration Not applicable. 6.8. Conflicting interests The authors declare that they have no competing interests. 6.9. Funding Maragheh University of Medical Sciences supported this re- search. 6.10. Author contributions Ideation and design: MY, and SA. Data collection: MY, SA, AB, AS, SRD. Analysis: MY, MA. Interpretation of results: All authors. Drafting the work: MY, SA, MA. Revising draft critically for important intellectual content: All authors. The authors read and approved the final manuscript. References 1. Gutierre MU, Telles JPM, Welling LC, Rabelo NN, Teix- eira MJ, Figueiredo EG. Biomarkers for traumatic brain injury: a short review. Neurosurg Rev. 2021;44(4):2091-7. 2. James SL, Theadom A, Ellenbogen RG, Bannick MS, Montjoy-Venning W, Lucchesi LR, et al. Global, regional, and national burden of traumatic brain injury and spinal cord injury, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 2019;18(1):56-87. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index 7 Archives of Academic Emergency Medicine. 2023; 11(1): e9 3. Lingsma HF, Yue JK, Maas AI, Steyerberg EW, Manley GT. Outcome prediction after mild and complicated mild traumatic brain injury: external validation of existing models and identification of new predictors using the TRACK-TBI pilot study. J Neurotrauma. 2015;32(2):83-94. 4. Bouvier D, Oris C, Brailova M, Durif J, Sapin V. Interest of blood biomarkers to predict lesions in medical imag- ing in the context of mild traumatic brain injury. Clin Biochem. 2020;85:5-11. 5. Calvi MR, Beretta L, Dell’Acqua A, Anzalone N, Licini G, Gemma M. Early prognosis after severe traumatic brain injury with minor or absent computed tomography scan lesions. J Trauma Acute Care Surg. 2011;70(2):447-51. 6. Laalo JP, Kurki TJ, Sonninen PH, Tenovuo OS. Relia- bility of diagnosis of traumatic brain injury by com- puted tomography in the acute phase. J Neurotrauma. 2009;26(12):2169-78. 7. Callahan MJ, MacDougall RD, Bixby SD, Voss SD, Robert- son RL, Cravero JP. Ionizing radiation from computed tomography versus anesthesia for magnetic resonance imaging in infants and children: patient safety consid- erations. Pediatr Radiol. 2018;48(1):21-30. 8. Isokuortti H, Iverson GL, Silverberg ND, Kataja A, Bran- der A, Öhman J, et al. Characterizing the type and loca- tion of intracranial abnormalities in mild traumatic brain injury. J Neurosurg. 2018;129(6):1588-97. 9. Amoo M, Henry J, O’Halloran PJ, Brennan P, Husien MB, Campbell M, et al. S100B, GFAP, UCH-L1 and NSE as predictors of abnormalities on CT imaging following mild traumatic brain injury: a systematic review and meta-analysis of diagnostic test accuracy. Neurosurg Rev. 2022;45(2):1171-93. 10. Nakhjavan-Shahraki B, Yousefifard M, Oraii A, Sarveazad A, Hosseini M. Meta-analysis of neuron specific enolase in predicting pediatric brain injury outcomes. EXCLI J. 2017;16:995. 11. Edalatfar M, Piri SM, Mehrabinejad M-M, Mousavi M-S, Meknatkhah S, Fattahi M-R, et al. Biofluid biomarkers in traumatic brain injury: a systematic scoping review. Neu- rocrit Care. 2021;35(2):559-72. 12. Wang KK, Munoz Pareja JC, Mondello S, Diaz-Arrastia R, Wellington C, Kenney K, et al. Blood-based traumatic brain injury biomarkers - Clinical utilities and regulatory pathways in the United States, Europe and Canada. Ex- pert Rev Mol Diagn. 2021;21(12):1303-21. 13. Mendoza D, López K, Echeverri R, Pastor L, Rueda S, Fer- nández Londoño L, et al. Utility of biomarkers in trau- matic brain injury: a narrative review. Revista Colom- biana de Anestesiologia. 2020;48:155-61. 14. Al-Adli N, Akbik OS, Rail B, Montgomery E, Caldwell C, Barrie U, et al. The Clinical Use of Serum Biomarkers in Traumatic Brain Injury: A Systematic Review Stratified by Injury Severity. World Neurosurg. 2021;155:e418-e38. 15. Seidenfaden S-C, Kjerulff JL, Juul N, Kirkegaard H, Møller MF, Münster A-MB, et al. Diagnostic accuracy of prehos- pital serum S100B and GFAP in patients with mild trau- matic brain injury: a prospective observational multi- center cohort study–“the PreTBI I study”. Scand J Trauma Resusc Emerg Med. 2021;29(1):75. 16. Middleton JL. UCH-L1 and GFAP Testing (i-STAT TBI Plasma) for the Detection of Intracranial Injury Follow- ing Mild Traumatic Brain Injury. Am Fam Physician. 2022;105(3):313-4. 17. Brooke BS, Schwartz TA, Pawlik TM. MOOSE Reporting Guidelines for Meta-analyses of Observational Studies. JAMA Surg. 2021;156(8):787-8. 18. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the qual- ity assessment of diagnostic accuracy studies. Ann Intern Med. 2011;155(8):529-36. 19. Guyatt GH, Oxman AD, Schünemann HJ, Tugwell P, Knot- tnerus A. GRADE guidelines: a new series of articles in the Journal of Clinical Epidemiology. J Clin Epidemiol. 2011;64(4):380-2. 20. Bazarian JJ, Biberthaler P, Welch RD, Lewis LM, Barzo P, Bogner-Flatz V, et al. Serum GFAP and UCH-L1 for pre- diction of absence of intracranial injuries on head CT (ALERT-TBI): a multicentre observational study. Lancet Neurol. 2018;17(9):782-9. 21. Çevik S, Özgenç MM, Güneyk A, Evran Ş, Akkaya E, Çalış F, et al. NRGN, S100B and GFAP levels are significantly in- creased in patients with structural lesions resulting from mild traumatic brain injuries. Clin Neurol Neurosurg. 2019;183:105380. 22. Clarke GJB, Skandsen T, Zetterberg H, Einarsen CE, Feyling C, Follestad T, et al. One-Year Prospective Study of Plasma Biomarkers From CNS in Patients With Mild Traumatic Brain Injury. Front Neurol. 2021;12:643743. 23. Gardner RC, Rubenstein R, Wang KKW, Korley FK, Yue JK, Yuh EL, et al. Age-Related Differences in Diagnos- tic Accuracy of Plasma Glial Fibrillary Acidic Protein and Tau for Identifying Acute Intracranial Trauma on Com- puted Tomography: A TRACK-TBI Study. J Neurotrauma. 2018;35(20):2341-50. 24. Lagerstedt L, Egea-Guerrero JJ, Bustamante A, Rodríguez-Rodríguez A, El Rahal A, Quintana-Diaz M, et al. Combining H-FABP and GFAP increases the capacity to differentiate between CT-positive and CT- negative patients with mild traumatic brain injury. PLoS One. 2018;13(7):e0200394. 25. Okonkwo DO, Puffer RC, Puccio AM, Yuh EL, Yue JK, Diaz-Arrastia R, et al. Point-of-Care Platform Blood Biomarker Testing of Glial Fibrillary Acidic Protein versus S100 Calcium-Binding Protein B for Prediction of Trau- This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index S. Ahmadi et al. 8 matic Brain Injuries: A Transforming Research and Clin- ical Knowledge in Traumatic Brain Injury Study. J Neuro- trauma. 2020;37(23):2460-7. 26. Papa L, Lewis LM, Falk JL, Zhang Z, Silvestri S, Gior- dano P, et al. Elevated levels of serum glial fibrillary acidic protein breakdown products in mild and moderate trau- matic brain injury are associated with intracranial le- sions and neurosurgical intervention. Ann Emerg Med. 2012;59(6):471-83. 27. Papa L, Ladde JG, O’Brien JF, Thundiyil JG, Tesar J, Leech S, et al. Evaluation of Glial and Neuronal Blood Biomark- ers Compared With Clinical Decision Rules in Assess- ing the Need for Computed Tomography in Patients With Mild Traumatic Brain Injury. JAMA Netw Open. 2022;5(3):e221302. 28. Posti JP, Takala RSK, Lagerstedt L, Dickens AM, Hossain I, Mohammadian M, et al. Correlation of Blood Biomarkers and Biomarker Panels with Traumatic Findings on Com- puted Tomography after Traumatic Brain Injury. J Neuro- trauma. 2019;36(14):2178-89. 29. Welch RD, Ayaz SI, Lewis LM, Unden J, Chen JY, Mika VH, et al. Ability of Serum Glial Fibrillary Acidic Pro- tein, Ubiquitin C-Terminal Hydrolase-L1, and S100B To Differentiate Normal and Abnormal Head Computed Tomography Findings in Patients with Suspected Mild or Moderate Traumatic Brain Injury. J Neurotrauma. 2016;33(2):203-14. 30. Yue JK, Yuh EL, Korley FK, Winkler EA, Sun X, Puffer RC, et al. Association between plasma GFAP concentrations and MRI abnormalities in patients with CT-negative trau- matic brain injury in the TRACK-TBI cohort: a prospec- tive multicentre study. Lancet Neurol. 2019;18(10):953- 61. 31. Safari S, Radfar F, Baratloo A. Thoracic injury rule out cri- teria and NEXUS chest in predicting the risk of traumatic intra-thoracic injuries: A diagnostic accuracy study. In- jury. 2018;49(5):959-62. 32. Meier TB, Huber DL, Bohorquez-Montoya L, Nitta ME, Savitz J, Teague TK, et al. A Prospective Study of Acute Blood-Based Biomarkers for Sport-Related Concussion. Ann Neurol. 2020;87(6):907-20. 33. Asken BM, Bauer RM, DeKosky ST, Svingos AM, Hro- mas G, Boone JK, et al. Concussion BASICS III: Serum biomarker changes following sport-related concussion. Neurology. 2018;91(23):e2133-e43. 34. Yuh EL, Cooper SR, Mukherjee P, Yue JK, Lingsma HF, Gordon WA, et al. Diffusion tensor imaging for outcome prediction in mild traumatic brain injury: a TRACK-TBI study. J Neurotrauma. 2014;31(17):1457-77. 35. Sharma R, Rosenberg A, Bennett ER, Laskowitz DT, Acheson SK. A blood-based biomarker panel to risk-stratify mild traumatic brain injury. PLoS One. 2017;12(3):e0173798. 36. Bogoslovsky T, Wilson D, Chen Y, Hanlon D, Gill J, Jeromin A, et al. Increases of Plasma Levels of Glial Fibril- lary Acidic Protein, Tau, and Amyloid β up to 90 Days af- ter Traumatic Brain Injury. J Neurotrauma. 2017;34(1):66- 73. 37. Thelin EP, Zeiler FA, Ercole A, Mondello S, Büki A, Bellan- der BM, et al. Serial Sampling of Serum Protein Biomark- ers for Monitoring Human Traumatic Brain Injury Dy- namics: A Systematic Review. Front Neurol. 2017;8:300. 38. Calcagnile O, Anell A, Undén J. The addition of S100B to guidelines for management of mild head injury is poten- tially cost saving. BMC Neurol. 2016;16(1):200. 39. Ruan S, Noyes K, Bazarian JJ. The economic impact of S- 100B as a pre-head CT screening test on emergency de- partment management of adult patients with mild trau- matic brain injury. J Neurotrauma. 2009;26(10):1655-64. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index 9 Archives of Academic Emergency Medicine. 2023; 11(1): e9 Table 1: Summary characteristics of included studies Author, year, country Study type GCS1 Sample size Mean Age (year) Male (n) Assay TBI criteria Timing (hrs) TBI patients (n) Non-TBI patients (n) Cut-offs (pg/ml) Bazarian, 2018, Multiple coun- tries Trial 9 to 15 1780 49.9 1107 ELISA CT <12 107 1673 22 Cevik, 2019, Turkey Cross- sectional 14 to 15 48 24 38 ELISA CT <4 24 24 230 Clarke, 2021, Norway Cohort 13 to 15 343 32.5 206 ELISA CT and MRI <24 76 267 57.5 Gardner, 2018, USA Cohort 13 to 15 169 41.88 127 ELISA CT <24 56 113 430 Lagerstedt, 2018, Spain and Switzerland Cohort 15 241 48.7 to 52.4 96 and 67 ELISA CT <6 38 203 97.3 Okonkwo, 2020, USA Trial 13 to 15 1137 40.12 435 ELISA CT <24 358 779 13.1, 37.8, 113.3, 190.1 Papa, 2012, USA Cohort 9 to 15 117 38 ELISA CT <4 32 85 35 Papa, 2022, USA Cohort 13 to 15 349 40 240 ELISA CT <4 23 326 67 Posti, 2019, Fin- land Cohort 13 to 15 93 56 37 ELISA CT <24 37 56 66.62, 132, and 540 Welch, 2016, Multiple coun- tries Cohort 9 to 15 251 45.6 151 ELISA CT 6 36 215 15 Yue, 2019, USA Cohort 13 to 15 450 36.3 285 ELISA MRI <24 120 330 4.4, 12.95, 25.15, 71.95, 282.7, and 848.75, 1- Data was presented as range. CT: Computed tomography scan; ELISA: enzyme-linked immunosorbent assay; GCS: Glasgow coma scale; N: Number; TBI: Traumatic brain injury; MRI: magnetic resonance imaging. Table 2: Risk of bias assessment of included studies based on QUADAS-2 tool Study Risk of bias Applicability Patients’ selection Index test Reference standard Flow and timing Patients’ selection Index test Reference standard Overall Bazarian, 2018 , , , , , , , , Cevik, 2019 , , , , , , , , Clarke, 2021 , , , , , , , , Gardner, 2018 , , , , , , , , Lagerstedt, 2018 , , , , , , , , Okonkwo, 2020 , , , , , , , , Papa, 2012 , , , , , , , , Papa, 2022 , , , , , , , , Posti, 2019 , , , , , , , , Welch, 2016 , , , , , , , , Yue, 2019 , , , , , , , , ,: Low risk This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index S. Ahmadi et al. 10 Table 3: Subgroup and sensitivity analyses for identification of source of heterogeneity among included studies Parameter No. of analyses Sensitivity P Specificity p2 I2(%) Study type Cohort 21 0.84 [0.79, 0.89] 0.03 0.42 [0.29, 0.55] 0.16 0.0 Cross-sectional 1 0.62 [0.25, 1.00] 0.14 0.39 [0.27, 0.50] 0.64 0.0 Trial 5 0.83 [0.73, 0.92] 0.01 0.27 [0.06, 0.47] 0.65 0.0 GFAP cut-offs (pg/ml) <100 17 0.83 [0.78, 0.89] <0.001 0.39 [0.24, 0.53] 0.60 0.0 100 to 300 5 0.79 [0.67, 0.90] <0.001 0.32 [0.08, 0.57] 0.94 0.0 >400 5 0.89 [0.81, 0.96] 0.12 0.43 [0.15, 0.71] 0.60 0.0 Control group CT negative patients 25 0.84 [0.79, 0.89] 0.16 0.40 [0.28, 0.52] 0.86 0.0 Healthy control 2 0.79 [0.61, 0.97] 0.15 0.22 [0.00, 0.52] 42 0.0 Severity of TBI Mild 24 0.80 [0.76, 0.84] <0.001 0.38 [0.25, 0.50] 0.93 82.0 Moderate to mild 3 0.95 [0.92, 0.99] 0.95 0.42 [0.08, 0.77] 0.74 82.0 GFAP: glial fibrillary acidic protein; CT: computed tomography scan; TBI: Traumatic brain injuries. Table 4: Certainty of evidence based on GRADE framework Outcome Sample size Diagnostic odds ratio Risk of bias Imprecision Inconsistency (I2) Indirectness Publication bias Judgment and level of evidence* Intracranial le- sion 4978 3.11 (95% CI: 1.69 to 5.72) Not serious Not present Not present Not present Not present Moderate: ⊕⊕⊕⊕⊕ Rated up one point: • Large magnitude of effect** *, based on our judgment **, based on diagnostic odds ratio Figure 1: Flow diagram of screening in current meta-analysis. GFAP: Glial fibrillary acidic protein; CT: Computed tomography scan; MRI: Magnetic resonance imaging; TBI: Traumatic brain injury. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index 11 Archives of Academic Emergency Medicine. 2023; 11(1): e9 Figure 2: Summary receiver operating characteristics (SROC) of glial fibrillary acidic protein )GFAP( in diagnosis of mild traumatic brain injuries. AUC: Area under the curve; SENS: Sensitivity; SPEC: Specificity. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index S. Ahmadi et al. 12 Figure 3: Sensitivity and Specificity of glial fibrillary acidic protein )GFAP( in diagnosis of mild traumatic brain injuries. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index 13 Archives of Academic Emergency Medicine. 2023; 11(1): e9 Figure 4: Diagnostic score and diagnostic odds ratio of glial fibrillary acidic protein )GFAP( in diagnosis of mild traumatic brain injuries. Figure 5: Publication bias in assessment of diagnostic accuracy of glial fibrillary acidic protein )GFAP( in detection of mild traumatic brain injuries. This open-access article distributed under the terms of the Creative Commons Attribution NonCommercial 3.0 License (CC BY-NC 3.0). Downloaded from: https://journals.sbmu.ac.ir/aaem/index.php/tem/index Introduction Methods Results Discussion Conclusions Declarations References