Comparing blood biomarkers to clinical decision rules to select patients suspected of traumatic brain injury for head computed tomography

Abstract

Introduction

Traumatic brain injury (TBI) is a major public health concern in the U.S. Recommendations for patients admitted in the emergency department (ED) to receive head computed tomography (CT) scan are currently guided by various clinical decision rules.

Objective

To compare how a blood biomarker approach compares with clinical decision rules in terms of predicting a positive head CT in adult patients suspected of TBI.

Methods

We retrospectively identified patients transported to our emergency department and underwent a noncontrast head CT due to suspicion of TBI and who had blood samples available. Published thresholds for serum and plasma glial fibrillary acidic protein (GFAP), ubiquitin carboxyl-terminal hydrolase-L1 (UCH-L1), and serum S100β were used to make CT recommendations. These blood biomarker-based recommendations were compared to those achieved under widely used clinical head CT decision rules (Canadian, New Orleans, NEXUS II, and ACEP Clinical Policy).

Results

Our study included 463 patients, of which 122 (26.3%) had one or more abnormalities presenting on head CT. Individual blood biomarkers achieved high negative predictive value (NPV) for abnormal head CT findings (88%–98%), although positive predictive value (PPV) was consistently low (25%–42%). A composite biomarker-based decision rule (GFAP+UCH-L1)’s NPV of 100% and PPV of 29% were comparable or better than those achieved under the clinical decision rules.

Conclusion

Blood biomarkers perform at least as well as clinical rules in terms of selecting TBI patients for head CT and may be easier to implement in the clinical setting. A prospective study is necessary to validate this approach.

Keywords

traumatic brain injury head computed tomography biomarkers plasma glial fibrillary acidic protein S100β ubiquitin carboxyl-terminal hydrolase-L1

Introduction

Traumatic brain injury (TBI) represents a major public health issue. Based on the 2016 Global Burden of Diseases, Injuries, and Risk Factors Study (GBD), there were 27 million new cases of TBI and over 55 million prevalent cases in the world.¹ In the United States alone, over 2.5 million annual emergency department (ED) visits are prompted by TBI, resulting in approximately 282,000 hospitalizations and 56,000 deaths.^2,3

A noncontrast head computerized tomography (CT) scan is typically the first imaging test used in the workup of patients suspected of TBI.⁴ However, only a fraction of head CTs in patients suspected of TBI will show positive findings for TBI,⁵ and different clinical decision rules have been developed to triage patients with suspected TBI and to maximize the yield of head CT in these patients. These rules include the Canadian CT Head Rule,^6,7 the New Orleans Criteria,^8,9 the National Emergency X-Radiography Utilization Study (NEXUS) II rule,¹⁰ and the American College of Emergency Physicians (ACEP) clinical policy.^11,12 While most TBI patients should receive the same head CT recommendation regardless of decision rule choice, there remains a fraction of patients for whom different rules will lead to conflicting decisions.^13,14 There is no consensus on which clinical rule should be used.¹⁵

Many recent studies have demonstrated the value of blood biomarkers in managing TBI. Among these blood biomarkers, S100β,¹⁶ neurofilament light polypeptide (NFL),¹⁷ glial fibrillary acidic protein (GFAP),¹⁸ and ubiquitin carboxyl-terminal hydrolase-L1 (UCH-L1)¹⁹ have been reported as most promising for diagnosing focal and diffuse brain injury, whereas neuron-specific enolase (NSE),²⁰ C-tau,²¹ and spectrin breakdown products (SBDPs)²² appear to be correlated with intracranial pressure.²³

The goal of this study was to compare the performance of clinical rules and blood biomarkers in terms of predicting a positive head CT in patients suspected of TBI.

Methods

Study population

We retrospectively enrolled patients transported to Stanford Health Care’s emergency department (ED) by ambulance or helicopter, for whom a trauma alert was triggered per established criteria (http://stan.md/2nw7pfD) and who underwent a noncontrast head CT scan due to suspicion of TBI, between December 2015 and April 2017. Stanford Hospital’s ED provides 24-h neurosurgery facilities and specialized intensive care to over 2500 TBI patients per year. Participant enrollment was limited to Mondays and Tuesdays due to research staff availability.

Patient eligibility was determined based on the following criteria: (1) at least 18 years old at admission, (2) presented to the ED with suspected TBI, and (3) have a blood draw as part of the standard of care. Non-English–speaking patients were excluded, as well as patients without the capacity to consent (including those with altered mental status and hearing impairments) if no legal authorized representative was available.

This study was approved by the Institutional Review Board at Stanford University. In accordance with the Helsinki Principles, research staff obtained written informed consent from patients or from their legal authorized representatives prior to enrollment.

Clinical information and clinical decision rules

Demographic and clinical characteristics were collected as part of their standard of care and extracted from our institution’s electronic health record system, including age, sex, mechanism of injury, Glasgow Comas Score at onset and at 2 h, clinical symptoms at presentation (alertness, general behavior, headache, vomiting, and seizure), alcohol use history, drug use history, and mortality status at hospital discharge. We also obtained records of other major extracranial traumatic injuries (OMEI) and other major intracranial non-traumatic injuries (OMII), for example, a stroke that would have caused the patient to get involved into a trauma.

We considered four commonly-used clinical decision rules for head CT in patients suspected of TBI—the Canadian CT Head Rule (Supplemental Table 1),^6,7 the New Orleans Criteria (Supplemental Table 2),^8,9 the NEXUS II rule (Supplemental Table 3),¹⁰ and ACEP Clinical Policy (Supplemental Table 4).^11,12 Based on each patient’s clinical characteristics, we then determined whether a head CT would be recommended under each rule.

Blood biomarkers

Approximately 25 mL of blood were drawn from and processed for each patient as part of their standard workup for trauma in the emergency room, and our IRB protocol allowed us to use any blood remaining after standard-of-care analyses. Blood samples were initially stored in ethylenediaminetetraacetic acid (EDTA) and serum-separating tubes (BD Vacutainer^® Part No. 367861 and 367986, Becton Dickinson, Heidelberg, Germany) in a lightly refrigerated manner at 0–8°C within 30 min of the blood draw, for the period of the standard-of-care blood analyses. A previous study showed that blood biomarkers in lightly refrigerated blood of TBI patients are stable for at least 3 days after blood draw.²⁴ Subsequently, blood was processed into plasma and serum sample using centrifugation (3500 revolutions per minute for 10 min) and stored as plasma and serum aliquot samples (USA Scientific 2.0 mL self-standing screw cap tube, Part No. 1420-9705 and 1420-9706, USA Scientific, Ocala, FL, USA) at −80°C (900 Series −86°C Upright Ultra-Low Temperature Freezer, ThermoFisher Scientific, Waltham, MA, USA) for batch processing. Samples were shipped overnight on dry ice to a laboratory (Banyan Biomarkers, San Diego, CA) for subsequent proteomic analyses. Technicians were blinded to clinical data and CT results. A sandwich enzyme-linked immunosorbent assay (ELISA) was utilized to detect GFAP, UCH-L1, and their breakdown products, carried out as previously described.²⁵ An electrochemiluminescence immunoassay (ECLIA) designed for in-vitro diagnostic testing (Roche, Cobas 6000) was used to detect S100β, measured in the serum only. The lower limit of quantification (LLoQ) for GFAP was 3 pg/mL, and the limit of detection (LoD) was 0.8 pg/mL. The LLoQ of UCH-L1 was 14 pg/mL and the LoD was 6 pg/mL. The LLoQ for S100β was 20 pg/mL and the LoD was 5 pg/mL. 10% of GPAF samples were below the LoD. Samples with signals greater than those in the quantification range were diluted and re-assayed.

Head CT imaging review

Noncontrast head CTs were independently reviewed for TBI imaging common data elements²⁶ as defined by the National Institute of Health (NIH). Two experienced neuroradiologists assessed the presence/absence of closed head injuries including skull fracture, pneumocephalus, hemorrhage, mass effect, and brain parenchymal injuries. The volumes of each type of hematoma or contusion, as well as the extent of midline shift, were quantified as continuous variables, while the extent of subarachnoid hemorrhage, intraventricular hemorrhage, brain edema/swelling, cisternal compression, and hydrocephalus was characterized on ordinal scales.

Statistical analysis

Descriptive statistics were used to characterize the cohort, with categorical characteristics summarized as counts/percentages and continuous characteristics summarized as means with standard deviation (SD) or as median with interquartile range (Q1, Q3) for skewed distributions. Demographic, clinical, and injury severity characteristics were stratified by presence/absence of head CT findings, whether the clinical decision rules recommended a head CT, or whether blood marker values were above a certain threshold value. Thresholds for GFAP (22 pg/mL) and UCH-L1 (327 pg/mL) were taken from Bazarian et al.’s multicenter study for the prediction of intracranial injuries on head CT,²⁷ and the threshold for S100β (105 pg/mL) was taken from Welch et al.’s study on differentiating normal and abnormal head CT findings in patients with suspected mild or moderate TBI.²⁵

We presented cross-tabulations of whether a head CT was recommended under each of the four head CT decision rules with actual head CT findings (abnormal vs normal) and estimated measures of test accuracy (sensitivity, specificity, positive predictive value [PPV], and negative predictive value [NPV]). These steps were then repeated for blood biomarker-based decision rules in both the full study cohort and in a subcohort without OMEI and OMII. For the clinical decision rules, the number of rules recommending a head CT was also summarized by head CT status.

Finally, we used the methodology as described in Leisenring et al.²⁸ to compare PPV and NPV, as estimated under the composite GFAP + UCH-L1 and the four clinical head CT decision rules. For each decision rule pair, a logistic regression model was fitted in the generalized estimating equations (GEE) framework to relate log odds of abnormal imaging status to decision rule, with records restricted to positive decision rule outcomes when comparing PPV and to negative decision rule outcomes when comparing NPV. Statistical significance was assessed at the 0.05 alpha level, and all analyses were performed in the R statistical computing framework, version 3.6.²⁹

We repeated our analyses three times: (1) for all the study patients together, (2) for the patients with a GCS of 15 exclusively, and (3) for the patients with a GCS of 13 or 14 exclusively.

Results

Study population

A total of 463 patients with suspected TBI (age: 50.8 ± 22.7 years; 38.2% female) were included in our study. The majority of patients (92.9%) were admitted due to fall from height or motor vehicular collision. The patients’ clinical characteristics are detailed in Table 1. Amongst our study patients, 333 (71.9%) had an admission GCS of 15, and 93 (20.1%) had an admission GCS of 13 or 14. All patients underwent a noncontrast head CT as part of their initial workup in the emergency room. Plasma and serum samples were available—and levels detectable—in 442 (95%) and 222 (48%) of the patients, respectively. Distribution of time from injury to ED admission/blood draw spanned 0 through 6 h with a median at 1 h.

Table 1.

Demographic, clinical, and injury severity characteristics of study cohort.

	Overall	Head CT findings		New Orleans Criteria		Plasma GFAP + UCH-L1
	Overall	Positive	Negative	Positive	Negative	Positive	Negative
No. patients	463	122	341	329	134	406	36
Age, mean ± SD, years	50.8 ± 22.7	61.3 ± 23.7	47.0 ± 21.2	52.6 ± 23.6	46.4 ± 19.7	51.7 ± 23.1	41.3 ± 17.4
Female sex, n (%)	177 (38.2)	52 (42.6)	125 (36.7)	117 (35.6)	60 (44.8)	148 (36.5)	19 (54.3)
Glasgow Coma Scale, median [Q1, Q3]	15 [14, 15]	15 [13, 15]	15 [15, 15]	15 [14, 15]	15 [15, 15]	15 [14, 15]	15 [15, 15]
Mechanism of injury, n (%)
Fall from height	180 (38.9)	75 (61.5)	105 (30.8)	145 (44.1)	35 (26.1)	158 (38.9)	12 (34.3)
Motor vehicular collision	250 (54.0)	45 (36.9)	205 (60.1)	161 (48.9)	89 (66.4)	220 (54.2)	18 (51.4)
Violence	25 (5.4)	2 (1.6)	23 (6.7)	17 (5.2)	8 (6.0)	22 (5.4)	3 (8.6)
Other/unknown	8 (1.7)	0 (0)	8 (2.3)	6 (1.8)	2 (1.5)	6 (1.5)	2 (5.7)
Other major intracranial nontraumatic injury, n (%)	39 (8.4)	4 (3.3)	35 (10.3)	35 (10.6)	4 (3.0)	29 (7.1)	6 (17.1)
Other major extracranial traumatic injury, n (%)	209 (45.1)	44 (36.1)	165 (48.4)	146 (44.4)	63 (47.0)	185 (45.6)	11 (31.4)
Secondary intracranial complications, n (%)	14 (3.0)	11 (9.0)	3 (0.9)	12 (3.6)	2 (1.5)	14 (3.4)	0 (0)
Secondary extracranial complications, n (%)	33 (7.1)	22 (18.0)	11 (3.2)	32 (9.7)	1 (0.7)	32 (7.9)	0 (0)
Follow-up brain imaging studies, n (%)
Any noncontrast head CTs	130 (28.1)	114 (93.4)	16 (4.7)	124 (37.7)	6 (4.5)	125 (30.8)	2 (5.7)
Any contrast-enhanced head CTs	14 (3.0)	13 (10.7)	1 (0.3)	14 (4.3)	0 (0)	14 (3.4)	0 (0)
Any intracranial and/or cervical CTAs	42 (9.1)	31 (25.4)	11 (3.2)	40 (12.2)	2 (1.5)	42 (10.3)	0 (0)
Any brain MRIs	26 (5.6)	17 (13.9)	9 (2.6)	25 (7.6)	1 (0.7)	23 (5.7)	1 (2.9)
Any intracranial and/or cervical MRAs	10 (2.2)	6 (4.9)	4 (1.2)	9 (2.7)	1 (0.7)	9 (2.2)	1 (2.9)
Headache, n (%)	91 (19.7)	26 (21.3)	65 (19.1)	65 (19.8)	26 (19.4)	74 (18.2)	12 (34.3)
Vomiting, n (%)	19 (4.1)	6 (4.9)	13 (3.8)	15 (4.6)	4 (3.0)	15 (3.7)	3 (8.6)
Seizure, n (%)	15 (3.2)	3 (2.5)	12 (3.5)	14 (4.3)	1 (0.7)	13 (3.2)	1 (2.9)
Alcohol use, n (%)	150 (32.5)	35 (28.7)	115 (33.9)	123 (37.5)	27 (20.3)	129 (31.9)	11 (31.4)
Blood concentration (mg/dL), median [Q1, Q3]	8.0 [1.0, 205]	4.0 [1.0, 159]	12.0 [1.0, 217]	12.0 [1.0, 226]	1.5 [1.0, 188]	12.0 [1.0, 204]	43.0 [1.2, 194]
Any head CT abnormalities, n (%)	122 (26.3)	122 (100)	0 (0)	119 (36.2)	3 (2.2)	119 (29.3)	0 (0)
Skull fracture, n (%)	31 (6.7)	31 (25.4)	0 (0)	31 (9.4)	0 (0)	30 (7.4)	0 (0)
Pneumocephalus, n (%)	10 (2.2)	10 (8.2)	0 (0)	10 (3.0)	0 (0)	10 (2.5)	0 (0)
Intracranial hemorrhage, n (%)	114 (24.6)	114 (93.4)	0 (0)	111 (33.7)	3 (2.2)	111 (27.3)	0 (0)
Mass effect, n (%)	24 (5.2)	24 (19.7)	0 (0)	24 (7.3)	0 (0)	24 (5.9)	0 (0)
Brain parenchymal injuries, n (%)	36 (7.8)	36 (29.5)	0 (0)	36 (10.9)	0 (0)	35 (8.6)	0 (0)

Head CT imaging findings

Among the 463 patients considered, head CT was abnormal in 122 (26.3%). Classes of abnormalities consisted of skull fracture (6.7% of 463), pneumocephalus (2.2%), intracranial hemorrhage (24.6%), mass effect (5.2%), and brain parenchymal injuries (7.8%).

Clinical decision rules

Table 2 reports—for each clinical decision rule and by abnormality status—how many patients for whom a head CT was recommended or not. When considering all study patients together, the resulting NPV was high across all clinical decision rules, ranging from 91% to 98%, while PPV ranged from 31% to 36%; the New Orleans Criteria achieved the highest NPV (98%) and PPV (36%) (Table 3). When considering only study patients with a GCS of 15, the NPV remained high while the PPV decreased, reflecting the lower frequency of positives cases in this subgroup of patients; again the New Orleans Criteria achieved the highest NPV (98%) and PPV (30%) (Table 3).

Table 2.

2x2 tables for clinical decision rules and head CT findings of TBI.

Head CT rules		Abnormal head CT	Normal head CT
Canadian	Head CT recommended	111	230
Canadian	Head CT not recommended	11	111
New Orleans	Head CT recommended	119	210
New Orleans	Head CT not recommended	3	131
NEXUS II	Head CT recommended	120	260
NEXUS II	Head CT not recommended	2	81
ACEP Clinical Policy	Head CT recommended	119	269
ACEP Clinical Policy	Head CT not recommended	3	72

Table 3.

Sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of four clinical decision rules for head CT findings of TBI.

Head CT rules
	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
All study patients (N = 463)
Canadian	91	33	33	91
New Orleans	98	38	36	98
NEXUS II	98	24	32	98
ACEP Clinical Policy	98	21	31	96
Study patients with GCS score of 15 (N = 333)
Canadian	85	38	25	91
New Orleans	97	46	30	98
NEXUS II	97	26	24	97
ACEP Clinical Policy	97	24	24	97
Study patients with GCS score of 13 or 14 (N = 93)
Canadian	97	15	37	90
New Orleans	97	15	37	90
NEXUS II	100	15	38	100
ACEP Clinical Policy	97	13	37	89

Consistency of clinical decision rules is summarized in Table 4. There were 241 (52%) patients for whom all four clinical decision rules recommended a head CT, but 146 (61%) of these were patients with no abnormalities and only 39% had an acute TBI injury detected on CT. The percentage of unnecessary head CTs is even higher when one to three clinical decision rules recommended a head CT. For the 341 patients with no abnormalities, a head CT was not recommended by any of the decision rules in only 21 (6.2%) cases. In other words, a normal patient would have received a CT recommendation under at least one of the rules 94% of the time.

Table 4.

Concordance of four clinical decision rules for head CT findings of TBI.

	Overall	Abnormal head CT	Normal head CT
Number of patients	463	122	341
No clinical decision rule recommended a head CT	21	0	21
1 out of 4 clinical decision rules recommended a head CT	48	5	43
2 out of 4 clinical decision rules recommended a head CT	54	5	49
3 out of 4 clinical decision rules recommended a head CT	99	17	82
All clinical decision rules recommended a head CT	241	95	146

While no patient with positive CT findings would have been missed if all four clinical decision rules were applied together, 14 patients with positive CT findings would have been missed by one, two, or three clinical decision rules. These 14 patients showed the following injuries on CT: skull fracture (1 patient), intracranial hemorrhage (11 patients), nonhemorrhagic contusions (5 patients), and mass effect (1 patient) (the total number of injuries is higher than the number of patients, because several patients showed multiple injuries).

Blood biomarkers

Table 5 reports the number of patients for whom a head CT was recommended or not by abnormal versus normal blood biomarker levels, using published cutoff values for each biomarker-based decision rule. The differences between plasma and serum results were not significant, and likely related to the different number of plasma and serum samples.

Table 5.

2x2 tables for blood biomarkers and head CT findings of TBI in the full study cohort.

Blood biomarker		Abnormal head CT	Normal head CT
Plasma GFAP	>22°pg/mL—Head CT recommended	115	159
Plasma GFAP	≤22°pg/mL—Head CT not recommended	4	162
Serum GFAP	>22°pg/mL—Head CT recommended	51	81
Serum GFAP	≤22°pg/mL—Head CT not recommended	4	85
Plasma UCH-L1	>327°pg/mL—Head CT recommended	113	266
Plasma UCH-L1	≤327°pg/mL—Head CT not recommended	6	57
Serum UCH-L1	>327 pg/mL—Head CT recommended	50	121
Serum UCH-L1	≤327°pg/mL—Head CT not recommended	6	45
Serum S100β	>105°pg/mL—Head CT recommended	41	120
Serum S100β	≤105°pg/mL—Head CT not recommended	3	25
Composite plasma biomarker	Plasma GFAP >22°pg/mL or UCH-L1 > 327—Head CT recommended	119	287
Composite plasma biomarker	Plasma GFAP <22°pg/mL and UCH-L1 < 327—Head CT not recommended	0	35
Composite serum biomarker	Serum GFAP >22°pg/mL or UCH-L1 > 327—Head CT recommended	56	137
Composite serum biomarker	Serum GFAP <22°pg/mL and UCH-L1 < 327—Head CT not recommended	0	29

When considering all study patients together, the NPV was high for individual and composite blood biomarkers, ranging from 88% to 100%, while the PPV ranged from 25 to 42%; the GFAP + UCH-L1 composite biomarker decision rule achieved the highest NPV (100%) and PPV (29%) (Table 6). When considering only study patients with a GCS of 15, the NPV remained high while the PPV decreased, reflecting the lower frequency of positive cases in this subgroup of patients; again the GFAP + UCH-L1 composite biomarker decision rule achieved the highest NPV (100%) amongst all blood biomarkers (Table 3).

Table 6.

Sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of blood biomarkers for head CT findings of TBI in the full study cohort.

Blood biomarker
	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
All study patients (N = 463)
Plasma GFAP >22 pg/mL	97	50	42	98
Serum GFAP >22 pg/mL	93	51	39	96
Plasma UCH-L1 >327 pg/mL	95	18	30	90
Serum UCH-L1 >327 pg/mL	89	27	29	88
Serum S100β >105 pg/mL	93	17	25	89
Plasma GFAP >22 pg/mL or UCH-L1 >327 pg/mL	100	11	29	100
Serum GFAP >22 pg/mL or UCH-L1 >327 pg/mL	100	17	29	100
Study patients with GCS score of 15 (N = 333)
Plasma GFAP >22 pg/mL	97	50	32	98
Serum GFAP >22 pg/mL	92	53	27	97
Plasma UCH-L1 >327 pg/mL	92	20	22	91
Serum UCH-L1 >327 pg/mL	88	32	20	93
Serum S100β >105 pg/mL	95	18	17	95
Plasma GFAP >22 pg/mL or UCH-L1 >327 pg/mL	100	12	22	100
Serum GFAP >22 pg/mL or UCH-L1 >327 pg/mL	100	20	20	100
Study patients with GCS score of 13 or 14 (N = 93)
Plasma GFAP >22 pg/mL	100	55	55	100
Serum GFAP >22 pg/mL	94	52	53	94
Plasma UCH-L1 >327 pg/mL	97	7	37	80
Serum UCH-L1 >327 pg/mL	88	7	35	50
Serum S100β >105 pg/mL	87	9	39	50
Plasma GFAP >22 pg/mL or UCH-L1 >327 pg/mL	100	5	37	100
Serum GFAP >22 pg/mL or UCH-L1 >327 pg/mL	100	7	38	100

Under the plasma GFAP + UCH-L1 composite biomarker decision rule, all patients with an abnormal head CT and 11% patients with a normal head CT would have been correctly triaged, respectively. In other words, none of the patients with TBI findings would have been missed for treatment, but 89% of normal patients would have been unnecessarily sent to the CT suite (Table 6).

Comparison of clinical decision rules and blood biomarkers

The NPV of the composite blood biomarker rule was significantly higher (p = 0.001) than that of the Canadian CT Head Rule, but was comparable to the other clinical decision rules in the study (p-values ranging from 0.077 to 0.152). Furthermore, the PPV of the composite plasma blood biomarker rule is similar to that of the ACEP Clinical Policy (p = 0.109), slightly lower than the Canadian CT Head Rule (p = 0.006) and the NEXUS II rule (p = 0.011), and significantly lower than the New Orleans Criteria (p < 0.001) (Table 7).

Table 7.

GEE-estimated^a p-values for comparing positive and negative predictive values between composite biomarker and head CT decision rules.

	PPV	NPV
Plasma GFAP + UCH-L1
Canadian	0.006	0.001
New Orleans	<0.001	0.080
NEXUS II	0.011	0.152
ACEP Clinical Policy	0.109	0.077

^aAs described in Leisenring et al.

Discussion

In recent years, blood biomarkers for TBI have been investigated in multiple studies. Blood concentrations of GFAP, tau, and neurofilament light polypeptide (NFL) were found to be higher in patients with mild TBI compared to those of controls.¹⁸ GFAP yielded an area under the curve (AUC) of 0.93, confirming its discriminatory power for distinguishing mild TBI from controls.¹⁸ S100β and NSE had areas under the curve of 0.63 and 0.64, respectively, for detecting clinically important brain injury.³⁰ For predicting delayed recovery after mild TBI, top blood biomarker candidates included calpain-derived aII-spectrin N-terminal fragment, tau A, NFL, and ghrelin.³¹

In this large prospective cohort, we primarily focused on GFAP, UCH-L1, and S100β. In our patients, plasma GFAP was shown to be the blood biomarker yielding the highest sensitivity (97%) and specificity (50%), which is consistent with previous studies.^18,19 A composite blood biomarker test result was best at detecting CT abnormality, also consistent with an earlier study showing that combining tau, NFL, and GFAP achieved satisfactory discrimination (AUC 0.80, 95% CI 0.69–0.90) for detecting MRI abnormalities, even in patients with mild TBI with a normal CT.¹⁸ This may be due that the different blood biomarkers peak at different times after the injury.³² Depending on how long after the trauma the patient is admitted and blood drawn, one blood biomarker may perform better than the others. Therefore, a composite blood biomarker panel is more likely to detect an abnormality depending on the time from injury to ED admission/blood draw.

Previous research has shown a low compliance with the clinical rules to select TBI patients for head CT, ranging around 15–60%.^33-36 Physicians are often either unaware of the clinical decision rules or ignore them in clinical practice due to their complexity.^34,36,37 In agreement with past literature,^6,9,38,39 our study found all four clinical decision rules to have excellent NPV but low PPV. We also noted that these clinical decision rules lacked consistency. Even the best clinical rule, the New Orleans Criteria, would have sent over 60% of the patients to the CT suite unnecessarily and missed 2.5% of patients with a positive head CT.

The NPV of the composite blood biomarker rule was significantly higher than that of the Canadian CT Head Rule, but was comparable to the other clinical decision rules in the study. Furthermore, the PPV of the composite plasma blood biomarker rule is similar to that of the ACEP Clinical Policy, slightly lower than the Canadian CT Head Rule and the NEXUS II rule, and significantly lower than the New Orleans Criteria.

Most of the clinical decision rules include a GCS threshold, reflecting the fact that they are meant to be used for patients with normal or almost normal GCS scores and that all patients with GCS less than 13 should undergo a head CT to assess for TBI. In our study, we conducted our analyses considering all patients together (including those with GCS less than 13 where a head CT would be recommended clinically) and also separately in patients with a normal GCS (15) and in patients with a GCS of 13–14. The results were the same in all three analyses, showing the high NPV and low PPV of clinical and biomarker rules.

We acknowledge several limitations to our study. All patients in our study underwent a head CT, which is indicative of more severe TBI that the clinical rules were originally developed for. Similarly, the rate of positive head CT (26%) is high, pointing to the same limitation. We tried to address this limitation by presenting a subanalysis focusing on patients with a GCS of 15, and another one for patients with a GCS of 13 or 14. Patients with blood samples may be victims of more severe TBI compared to those who did not get their blood drawn, and our cohort may not be representative of the general population admitted in the emergency room with a suspicion of TBI. We only tested three blood biomarkers. We applied the clinical decision rules retrospectively instead of these rules being applied by the clinicians treating the patients. Similarly, the blood biomarkers analysis was applied retrospectively. Finally, we did not compare the clinical decision rules to each other, only the blood biomarkers to the clinical decision rules.

In conclusion, our study shows that blood biomarkers perform at least as well as clinical rules in terms of selecting TBI patients for head CT and may be easier to implement in the clinical setting. A prospective study is necessary to validate this approach, which could have a significant impact in terms of reducing the number of unnecessary head CTs in patients’ victims of a trauma. Such a prospective study, and the use of blood biomarkers in general for patient screening, will be facilitated by the fact that GFAP and UCH-L1 are now available as point-of-care assay and can be obtained at the bedside.

Supplemental Material

Supplemental Material - Comparing blood biomarkers to clinical decision rules to select patients suspected of traumatic brain injury for head computed tomography

Supplemental Material for Comparing blood biomarkers to clinical decision rules to select patients suspected of traumatic brain injury for head computed tomography by Ying Li , Victoria Y Ding , Hui Chen , Guangming Zhu , Bin Jiang , Derek Boothroyd , Paymon G Rezaii , Anthony M Bet , Amy D Paulino , Art Weber , Olena Y Glushakova , Ronald L Hayes and Max Wintermark in The Neuroradiology Journal.

Footnotes

Author Contributions

1. Li: Methodology, Data collection, Data analysis, Manuscript drafting, and Manuscript editing. 2. Ding: Statistical Analysis, Manuscript drafting, and Manuscript editing. 3. Chen: Data collection, Data analysis, and Manuscript editing. 4. Zhu: Data collection, Data analysis, and Manuscript editing. 5. Jiang: Data collection, Data analysis, and Manuscript editing. 6. Boothroyd: Statistical Analysis and Manuscript editing. 7. Rezaii: Data collection, Data analysis, and Manuscript editing. 8. Bet: Data collection, Data analysis, and Manuscript editing. 9. Paulino: Methodology, Data analysis, and Manuscript editing. 10. Weber: Methodology, Data analysis, Manuscript editing. 11. Glushakova: Methodology, Data analysis, Manuscript editing. 12. Hayes: Study Design, Methodology, and Manuscript editing. 13. Wintermark: Study design, Project administration, Methodology, Manuscript drafting, and Manuscript editing.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Paymon G Rezaii

Max Wintermark

Supplemental Material

Supplemental material for this article is available online.

References

Global, regional, and national burden of neurological disorders, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol 2019; 18(5): 459–480.

Marin

Weaver

Yealy

, et al. Trends in visits for traumatic brain injury to emergency departments in the United States. JAMA 2014; 311(18): 1917–1919.

Taylor

Bell

Breiding

, et al. Traumatic Brain Injury-related emergency department visits, hospitalizations, and deaths - United States, 2007 and 2013. MMWR Surveill Summ 2017; 66(9): 1–16.

Wintermark

Sanelli

Anzai

, et al. Imaging evidence and recommendations for traumatic brain injury: conventional neuroimaging techniques. J Am Coll Radiol 2015; 12(2): e1–14.

Iverson

Lovell

Smith

, et al. Prevalence of abnormal CT-scans following mild head injury. Brain Inj 2000; 14(12): 1057–1061.

Davey

Saul

Russel

, et al. Application of the Canadian computed tomography head rule to patients with minimal head injury. Ann Emerg Med 2018; 72(4): 342–350.

Stiell

Lesiuk

Wells

, et al. The Canadian CT Head Rule Study for patients with minor head injury: rationale, objectives, and methodology for phase I (derivation). Ann Emerg Med 2001; 38(2): 160–169.

Papa

Stiell

Clement

, et al. Performance of the Canadian CT Head Rule and the New Orleans Criteria for predicting any traumatic intracranial injury on computed tomography in a United States Level I trauma center. Acad Emerg Med 2012; 19(1): 2–10.

Smits

Dippel

de Haan

, et al. External validation of the Canadian CT Head Rule and the New Orleans Criteria for CT scanning in patients with minor head injury. JAMA 2005; 294(12): 1519–1525.

10.

Mower

Hoffman

Herbert

, et al. Developing a clinical decision instrument to rule out intracranial injuries in patients with minor head trauma: methodology of the NEXUS II investigation. Ann Emerg Med 2002; 40(5): 505–514.

11.

DeAngelis

Lou

, et al. Head CT for minor head injury presenting to the emergency department in the era of choosing wisely. West J Emerg Med 2017; 18(5): 821–829.

12.

Manquen

Combs

Mazur-Mosiewicz

, et al. A review of research efforts to address the 2008 ACEP guideline for mild traumatic brain injury. Am J Emerg Med 2019; 37(1): 73–79.

13.

Harnan

Pickering

Pandor

, et al. Clinical decision rules for adults with minor head injury: a systematic review. J Trauma 2011; 71(1): 245–251.

14.

Easter

Haukoos

Meehan

, et al. Will neuroimaging reveal a severe intracranial injury in this adult with minor head trauma?: The rational clinical examination systematic review. JAMA 2015; 314(24): 2672–2681.

15.

Foks

van den Brand

Lingsma

, et al. External validation of computed tomography decision rules for minor head injury: prospective, multicentre cohort study in the Netherlands. BMJ 2018; 362: k3527.

16.

Nygren De Boussard

Fredman

Lundin

, et al. S100 in mild traumatic brain injury. Brain Inj 2004; 18(7): 671–683.

17.

Iverson

Reddi

Posti

, et al. Serum neurofilament light is elevated differentially in older adults with uncomplicated mild traumatic brain injuries. J Neurotrauma 2019; 36(16): 2400–2406.

18.

Gill

Latour

Diaz-Arrastia

, et al. Glial fibrillary acidic protein elevations relate to neuroimaging abnormalities after mild TBI. Neurology 2018; 91(15): e1385–e1389.

19.

Diaz-Arrastia

Wang

Papa

, et al. Acute biomarkers of traumatic brain injury: relationship between plasma levels of ubiquitin C-terminal hydrolase-L1 and glial fibrillary acidic protein. J Neurotrauma 2014; 31(1): 19–25.

20.

Thelin

Jeppsson

Frostell

, et al. Utility of neuron-specific enolase in traumatic brain injury; relations to S100B levels, outcome, and extracranial injury severity. Crit Care 2016; 20: 285.

21.

Forouzan

Motamed

Delirrooyfard

, et al. Serum cleaved tau protein and clinical outcome in patients with minor head trauma. Open Access Emerg Med 2020; 12: 7–12.

22.

Chen

Shi

Zheng

, et al. Role of α-II-spectrin breakdown products in the prediction of the severity and clinical outcome of acute traumatic brain injury. Exp Ther Med 2016; 11(5): 2049–2053.

23.

Yokobori

Hosein

Burks

, et al. Biomarkers for the clinical differential diagnosis in traumatic brain injury--a systematic review. CNS Neurosci Ther 2013; 19(8): 556–565.

24.

Rezaii

Grant

Zeineh

, et al. Stability of blood biomarkers of traumatic brain injury. J Neurotrauma 2019; 36(16): 2407–2416.

25.

Welch

Ayaz

Lewis

, et al. Ability of serum glial fibrillary acidic protein, 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): 203–214.

26.

Haacke

Duhaime

Gean

, et al. Common data elements in radiologic imaging of traumatic brain injury. J Magn Reson Imaging 2010; 32(3): 516–543.

27.

Bazarian

Biberthaler

Welch

, et al. Serum GFAP and UCH-L1 for prediction of absence of intracranial injuries on head CT (ALERT-TBI): a multicentre observational study. Lancet Neurol 2018; 17(9): 782–789.

28.

Leisenring

Alonzo

Pepe

. Comparisons of predictive values of binary medical diagnostic tests for paired designs. Biometrics 2000; 56(2): 345–351.

29.

R Core Team . R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org/. Published 2019 (2019, Accessed May 2020). Published

30.

Wolf

Frantal

Pajenda

, et al. Predictive value of neuromarkers supported by a set of clinical criteria in patients with mild traumatic brain injury: S100B protein and neuron-specific enolase on trial: clinical article. J Neurosurg 2013; 118(6): 1298–1303.

31.

Gan

Stein

Swanson

, et al. Blood biomarkers for Traumatic Brain Injury: a quantitative assessment of diagnostic and prognostic accuracy. Front Neurol 2019; 10: 446.

32.

Papa

Brophy

Welch

, et al. Time course and diagnostic accuracy of glial and neuronal blood biomarkers GFAP and UCH-L1 in a large cohort of trauma patients with and without mild traumatic brain injury. JAMA Neurol 2016; 73(5): 551–560.

33.

Tan

Lim

AME

Ong

, et al.

Computed tomography of the head for adult patients with minor head injury: are clinical decision rules a necessary evil?

Singapore Med J 2018; 59(4): 199–204.

34.

Melnick

Szlezak

Bentley

, et al. CT overuse for mild traumatic brain injury. Jt Comm J Qual Patient Saf 2012; 38(11): 483–489.

35.

Korley

Morton

Hill

, et al. Agreement between routine emergency department care and clinical decision support recommended care in patients evaluated for mild traumatic brain injury. Acad Emerg Med 2013; 20(5): 463–469.

36.

Eagles

Stiell

Clement

, et al. International survey of emergency physicians' awareness and use of the Canadian Cervical-Spine Rule and the Canadian Computed Tomography Head Rule. Acad Emerg Med 2008; 15(12): 1256–1261.

37.

Heskestad

Baardsen

Helseth

, et al. Guideline compliance in management of minimal, mild, and moderate head injury: high frequency of noncompliance among individual physicians despite strong guideline support from clinical leaders. J Trauma 2008; 65(6): 1309–1313.

38.

Schachar

Zampolin

Miller

, et al. External validation of the New Orleans Criteria (NOC), the Canadian CT Head Rule (CCHR) and the National Emergency X-Radiography Utilization Study II (NEXUS II) for CT scanning in pediatric patients with minor head injury in a non-trauma center. Pediatr Radiol 2011; 41(8): 971–979.

39.

Bouida

Marghli

Souissi

, et al. Prediction value of the Canadian CT head rule and the New Orleans criteria for positive head CT scan and acute neurosurgical procedures in minor head trauma: a multicenter external validation study. Ann Emerg Med 2013; 61(5): 521–527.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.32 MB