Serum cleaved tau protein levels and clinical outcome in adult patients with closed head injury☆☆☆★

Introduction 

There is increased interest in the utility of chemical markers in patients with closed head injury (CHI).1, 2, 3, 4, 5 This is motivated by the need to improve diagnosis and prognostication and by the recognition that patients with head injuries may suffer long-term neurocognitive impairments.6, 7 Assessment currently relies on indirect measures such as the Glasgow Coma Scale (GCS) and imaging studies. The usefulness of these tests to establish prognosis is limited.8 It would be beneficial to supplement these measures with a more direct marker of brain injury.

Tau (τ) is a microtubule-associated structural protein localized to the axons of central nervous system (CNS) neurons. It is released from CNS neurons into the extracellular space when an axonal injury such as trauma or hypoxia occurs. After release, it is proteolytically cleaved at the N- and C-terminals and diffuses into the cerebrospinal fluid (CSF) and plasma. Monoclonal antibodies that recognize this cleaved form of τ protein have been used in an enzyme-linked immunosorbent assay (ELISA) format to quantify CSF and plasma levels of the cleaved τ protein (τc).9 CSF τc levels were elevated more than 1,000 times in patients who had a CHI when compared with normal and neurologically diseased controls.9, 10 In the non-neurologically diseased controls, there were no individuals with a CSF τc level of more than 0.

The objective of this preliminary pilot study was to determine the relationship between the presence of serum τc, ICI, and outcome at hospital discharge in adult patients with CHI.

Materials and methods 

This was a prospective pilot study of adults with CHI presenting to an urban Level I trauma center. A convenience sample of patients, as determined by research assistant availability, was enrolled from June 1998 through July 1999. The study protocol was approved by the hospital's institutional review board.

Patients were eligible for enrollment if, in the opinion of the treating emergency physician, they were suspected of having sustained a CHI, presented within 10 hours of their injury, and were going to receive a head computed tomographic (CT) scan as part of their emergency department evaluation. Head CT findings were not a factor in eligibility and were performed after patient enrollment.

Demographic information was recorded at the time of enrollment. The GCS score, as determined by the initial treating physician, was used in this study. If the patient was transferred from an outlying institution, that institution's initial GCS score was used. The mechanism of injury was determined from the treating physician's record. Injury mechanism categories were falls, assaults, motor vehicle crashes, and pedestrians struck by vehicles. Time of injury was determined from emergency medical services (EMS) records of the incident or the initial treating physician's record. The ED readings of the blood pressure and oxygen saturation during the patient's treatment were recorded. Associated injuries were retrospectively determined from the medical record. All recorded injuries were included, with the exception of superficial lacerations.

Blood samples were obtained on presentation. Samples were drawn into 10-mL serum separator tubes and centrifuged at 13,000g for 15 minutes after collection. The serum samples were then frozen at −70°C (-94°F) and later assayed for τc protein using a sandwich ELISA technique previously described elsewhere.9 The sensitivity of the ELISA assay is 0.1 ng/mL. This technique yielded a value for the concentration of the τc; however, only the presence or absence of τc is used in this study. The detailed pharmacokinetics of τc are not known and, therefore, drawing conclusions on the basis of a numeric τc level is not justified.

The final interpretation of the patient's initial head CT, as determined by an attending radiologist, was used for the results. Patients with any intracranial injury were assigned to the ICI group. Those patients with normal head CT findings and those with an isolated skull fracture were assigned to the NICI group.

Medical record review was used to determine the outcome at hospital discharge. Patients were considered to have a “good” outcome if they were discharged to home. The outcome was defined as “poor” if they were transferred to a nursing home or died while in hospital.

Statistical analysis of the data was performed using SAS version 8.1 (SAS, Cary, NC). Student's t test (2-tailed) and Fisher's exact test were used as appropriate. A value of P less than or equal to .05 was considered statistically significant. The 95% confidence intervals (CIs) were calculated for odds ratios when appropriate.

Results 

Twenty-eight patients were enrolled during the study period of June 1998 through July 1999. Twenty-three of these patients were admitted, and 5 were discharged home from the ED. For purposes of comparison, there were 1,664 patients admitted for traumatic injury during the same time interval, and 606 patients underwent head CT scanning. Head CT abnormalities were found in 19 (68%) of the 28 patients. Seventeen (61%) patients had ICIs, 2 (7%) had isolated skull fractures, and 9 (32%) had normal CTs. As a result, 17 patients were dichotomized into the ICI group and 11 into the NICI group. In the ICI group, 5 (29%) patients were transferred from outlying hospitals, and the remaining 12 (71%) were brought by EMS to the ED from the scene of their injury. Two (18%) patients in the NICI group were transferred from outlying hospitals, and the remaining 9 (82%) arrived by EMS from the scene of injury.

Table 1 shows the demographics, time from injury to τc sample, initial GCS scores, and mechanism of injury for the ICI and NICI groups.

Table 1. Demographics of ICI and NICI groups.

	Demographics	ICI (n=17)	NICI (n=11)	P Value
	Men/women	13/4	9/2	1*
	Black/white	2/15	0/11	.51*
	Age, y (mean±SD)	43±16	36±15	.65†
	Initial GCS score (mean±SD)	10±5	12±4	.16†
	Time of τc sample from time of injury, h (mean±SD)	3.9±3.0	2.4±2.3	.11†
	Motor vehicle crash, no./%	9/53	6/55
	Falls, no./%	6/35	2/18
	Assault, no./%	0/0	2/18
	Pedestrians struck, no./%	2/12	1/9
	*Fisher's exact test (2-tailed). †Student's t test (2-tailed).

Ten (59%) patients in the ICI group had a poor outcome. Four of the 10 patients died while in the hospital; the other 6 required nursing home placement at discharge. All deaths were caused by the patients' neurologic injuries and not by their associated injuries. Only 1 (9%) patient in the NICI group had a poor outcome requiring nursing home placement resulting from pain control for associated pelvic fractures and not for neurologic problems. There were no deaths in the NICI group. The patient with the C2 fracture in the ICI group had a serum τc level of 0 but did have a subdural hematoma evident on the initial head CT scan. This patient required nursing home placement.

Table 2 shows patient outcome and ICI as a function of serum τc.

Table 2. ICI and outcome versus serum τclevels.

	Serum τc	ICI*	NICI	Bad Outcome†	Good Outcome
	τc>0	9	1	7	3
	τc=0	8	10	4	14
	*P =.02; Fisher's exact test (2-tailed). †P =.017; Fisher's exact test (2-tailed).

Discussion 

This pilot study demonstrated that a serum τc level of more than 0 was associated with a significant chance of an ICI in the initial head CT scan and an increased likelihood of a poor outcome in the adult CHI population. To our knowledge, this is the first study of τc in this population. These results suggest that τc could be a useful marker for determining the presence or absence of intracranial injury in the ED patient with a head injury. Serum τ may also contain information useful in estimating long-term prognosis and outcome both in the ED and the ICU.

Others have examined the utility of CNS injury markers for patients with CHI, such as S-100. S-100 is a calcium-binding protein localized to the astroglial cells of the CNS. Woertgen et al8 found that a serum S-100β level of more than 2 μg/mL had a positive predictive value (PPV) and negative predictive value (NPV) of 87% and 77%, respectively, for the outcome of either death or severe to moderate disability at 11 months after injury. Their blood samples were drawn at an average of 2.6 hours after injury (range 1 to 6 hours), which is close to the mean time to τ sample of 3.2 hours in the current study. Their outcome measure is roughly equivalent to a poor outcome in this study, and a similar analysis of our data using τc level of more than 0 yields a PPV and NPV of 70% (95% CI 41% to 90%) and 78% (95% CI 62% to 89%), respectively, for poor outcome. Our outcome determination differed from their work in that it was determined at hospital discharge and no long-term follow-up of patients was done.

Yamazaki et al4 examined the utility of serum neuron-specific enolase (NSE) and myelin basic protein (MBP) in patients with head injury. They found that an initial NSE of more than 20 ng/mL had a sensitivity and specificity of 87% and 60%, respectively, for predicting death in their patients. An MBP of more than 4 ng/mL was prognostic of death in these patients with a sensitivity and specificity of 87% and 100%, respectively. This can be compared with a sensitivity and specificity of 64% and 82%, respectively, using a τc level of more than 0 as a marker for poor outcome. Although these comparisons are not exact, it is compelling that the results presented here are qualitatively similar to those found in other outcome studies using CNS injury markers in the CHI population.

Three major limitations of this study are the small number of participants, selection bias, and the lack of detailed out-of-hospital information. The low patient number results in low statistical power for this study. The selection bias likely has lead to a more severely injured patient population being enrolled as participants and, therefore, a greater likelihood of observing an effect. The bias is caused by 2 factors. First, this study was done in an urban Level I trauma center, and more seriously ill patients can be expected to present to a trauma center than to a non-tertiary care center. Second, the study facility is also a referral center, and 7 patients enrolled in this study were transferred from other facilities. Further, there is little information available regarding the out-of-hospital course of these patients, and therefore no comparison of the out-of-hospital course between the ICI and NICI groups can be performed. It is possible that these groups may substantially differ in their out-of-hospital course and EMS interventions. As a result of the aforementioned limitations, the conclusions drawn here are preliminary in nature. Certainly, there remains much work to be done in evaluating the potential utility of τc as a test available to the emergency physician in the evaluation and treatment of the adult patient with CHI.

This was the first study to examine the utility of a τc protein assay in adult patients with CHI presenting to the ED. A τc level of more than 0 was associated with a poor outcome at hospital discharge and with an increased chance of an ICI as evidenced in the initial head CT. These results are encouraging and warrant further investigation in a larger prospective study.