JOURNAL OF NEUROTRAUMAVolume 17, Number 8, 2000Mary Ann Liebert, Inc.

Traumatic Brain Damage: Serum S-100 Protein MeasurementsRelated to Neuroradiological Findings

BERTIL ROMNER,1 TOR INGEBRIGTSEN,2 POUL KONGSTAD,1

and SVEND ERIK BØRGESEN3

ABSTRACT

This study was designed to investigate the correlation between S-100 protein serum measurementsand neuroradiological findings in patients with head injury. We studied 278 patients with minor,moderate, and severe head injuries and 110 controls with no history of neurological disease. Thestudy recruited patients from three Scandinavian neurotrauma centers. Serum levels of S-100 pro-tein were measured at admittance, and computed tomographic scans of the brain were obtainedwithin 24 h postinjury in all patients. In a subgroup of 45 patients with minor head injuries, mag-netic resonance imaging was also performed. Increased serum level of S-100 protein was detectedin 108 (39%) patients, and CT scan demonstrated intracranial pathology in 25 (9%) (brain contu-sion n 5 13, subdural hematoma n 5 6, epidural hematoma n 5 2, traumatic subarachnoid hemor-rhage n 5 2, and brain edema n 5 2). The proportion of patients with detectable serum level wassignificantly (p , 0.01) higher among those with intracranial pathology (92%) compared to thosewithout (34%). The negative predictive value of an undetectable S-100 serum level was 0.99. Un-detectable serum level of S-100 protein predicts normal intracranial findings on CT scan. Determi-nation of S-100 protein in serum may be used to select patients for CT scanning.

Key words: computed tomography; head injury; magnetic resonance imaging; management; S-100 pro-tein

641

1Department of Neurosurgery, University Hospital of Lund, Lund, Sweden.2Department of Neurosurgery, University Hospital of Tromsø, Tromsø, Norway.3Department of Neurosurgery, Rigshospitalet, Copenhagen, Denmark.

INTRODUCTION

TH E DIAG N OSIS O F TRA UM A TIC BRAIN INJURY (TBI) af-ter head injury is difficult to establish since specific

measures of the presence and severity of such injury areunavailable. The Mild Traumatic Brain Injury Commit-tee of the Head Injury Interdisciplinary Special InterestGroup of the American Congress of Rehabilitation Med-icine (1993) has suggested a clinical definition of mildTBI. Computed tomography (CT) or magnetic resonance

imaging (MRI) may demonstrate structural damage to thebrain (Culotta et al., 1996; Gandy et al., 1984; Gomez etal., 1996; Han et al., 1984; Hsiang et al., 1997; Inge-brigtsen et al., 1997; Jenkins et al., 1986; Levin et al.,1987; Shackford et al., 1992; Stein, 1996), but, in mostpatients with minor head injury, the presence of TBI re-mains unconfirmed.

Protein synthetized in astroglial cells or neurons havebeen proposed as markers of cell damage in the centralnervous system. The most common include fibrillary

acidic protein, myelin basic protein, creatine kinaseizoenzyme BB, neuron-specific enolase, and S-100 pro-tein. The S-100 protein is a small dimeric cytosolic pro-tein (Mr 5 22,000) and exists in various forms, depend-ing on its alfa or beta chain structure. The b b -form ishighly specific for tissue of the nervous system. S-100protein is metabolized in the kidney and excreted in theurine with a biological half-life of 2 h (Blomquist et al.,1997; Usui et al., 1989). It can be measured in arterial orvenous serum, is not affected by hemolysis, and remainsstable for hours without need for immediate centrifuga-tion and freezing for the sample.

Since 1993, we have determined S-100 protein levelsin serum after head injury and related the measurementsto CT and MRI findings, postconcussion symptoms, andneurobehavioral outcome (Ingebrigtsen et al., 1995,1997, 1999; Waterloo et al., 1997). The present reportsummarizes our total experience with S-100 proteinserum measurements with focus on their correlation toneuroradiological findings.

MATERIALS AND METHODS

Subjects

This prospective study recruited patients from threecenters in Scandinavia (Department of Neurosurgery,University Hospital of Tromsø, Norway; Department ofNeurosurgery, University Hospital of Lund, Sweden; andDepartment of Neurosurgery, Rigshospitalet, Copen-hagen, Denmark). The study was approved by the ethicscommittee at the University of Tromsø. We studied hos-pital admitted patients with mild, moderate, and severehead injury during a 5-year period (January 1993 to De-cember 1997).

The following criteria were required for inclusion inthe study group: (1) head injury with loss of conscious-ness (LOC), (2) blood sample for S-100 analysis collectedwithin 24 h after injury, and (3) CT scan performed within

24 h after the injury. LOC was considered to have oc-curred when the patient had amnesia for the trauma eventand if accompanying persons reported LOC. Patients witha history of neurological disease (e.g., multiple sclerosis,cerebral tumor, epilepsy, stroke, serious head injury)were excluded.

A total of 278 (Tromsø n 5 172, Lund n 5 76, Copen-hagen n 5 30) patients were included. They were 175(63%) men and 103 (37%) women, with a mean age of32 (range, 1–84) years. The injuries were caused by roadtraffic accidents in 122 (44%), falls in 80 (29%), sportsinjuries in 36 (13%), assaults in 30 (11%), and othercauses in 7 (3%). Table 1 shows the distribution of headinjury severity according to the Head Injury SeverityScale (HISS) (Stein and Spetell, 1995).

Seventy-five males and 35 females with no actual orprevious history of neurological disease were selected ascontrols. CT scan was not performed in these 110 indi-viduals.

Clinical and Radiological Examination

Neurological examination and assessment of the levelof consciousness according to the Glasgow Coma Scale(GCS) was performed in the emergency room. The headinjuries were classified according to the HISS as eithersevere (GCS 3–8), moderate (GCS 9–13), or mild (GCS14–15). Patients with severe head injury (GCS # 9) wereintubated and ventilated in the emergency room beforethe initial CT scan. All patients were admitted for at leastovernight observation.

The study protocol included nonenhanced CT scans ofthe brain and cranium in all patients. In a subgroup of 45patients with mild head injury (GCS 14–15, LOC for , 20min, absence of focal neurological deficits, and no signsof acute intracranial abnormality revealed by a CT scan),MRI was also performed. Images were obtained with aGyroscan T5 II 0.5-T magnet (Philips Medical Systems,Eindhoven, The Netherlands). The examination includedT1-weighted (374/15), T2-weighted spin echo (2,229/

ROMNER ET AL.

642

TABLE 1. DISTRIBUTION O F HEAD INJUR Y SEVERITY AC CO RDIN G TO TH E HEAD IN JURY SEV ER ITY SCALE , PRO P OR TION

OF S-100–PO SITIVE PA TIENTS IN EA CH SEV ERITY CATEG O RY A ND MEAN S-100 SERUM LEVELS

No. of patients in Mean S-100 serumHISS category No. of patients S-100–positive group level ( m g/L)

Mild 254 88 (35%)1 0.6a

Moderate 16 12 (75%)1 0.7a

Severe 8 8 (100%) 3.6a

ap 5 0.0001 (ANOVA analysis).HISS, Head Injury Severity Scale (Stein and Spetell, 1995).

20/90), and fluid attenuated inversion recovery (11,000/140) sequences of the brain.

For the purpose of statistical analysis, patients withmild head injury (GCS 14–15) were further divided intothe following subgroups: mild A, normal neuroradiolog-ical findings (n 5 228); mild B, CT-verified skull frac-ture and normal intracranial findings (n 5 14); mild C,normal intracranial CT, but cerebral contusions revealedby MRI (n 5 5); and mild D, CT-verified intracranialpathology (n 5 12).

S-100 Protein Analysis

A serum sample for S-100 analysis was drawn imme-diately (mean 3.8 h after injury; range, 0.5–24.0 h) afteradmission to the emergency room in the head-injured pa-tients. In the control group, serum was taken from a ve-nous cannula. Serum concentrations of S-100 b proteinwere analyzed with a sensitive, commercially availableimmunoradiometric assay kit (detection limit, 0.2 m g/L;Sangtec Medical, Bromma, Sweden). The methods havepreviously been described in detail (Ingebrigtsen et al.,1997; Nygaard et al., 1997). For the purpose of statisti-cal analysis, the head-injured patients were dichotomizedinto those with nondetectable serum levels (S-100 nega-tive group) and those with a serum level of at least 0.2m g/L (S-100 positive group).

Statistics

All data were entered into a common database(Tromsø). Statistical analysis were performed withStatView SE 1 Graphics (Abacus Concepts, Berkeley,CA). We compared means in multiple groups with one-way analysis of variance (ANOVA). All p values reportedare two-tailed. Comparison of proportions was performedwith the x 2 test for trends, including the continuity cor-rection. Relation between continuous variables and S-100protein levels were assessed with regression analysis. pvalues of # 0.05 were considered significant.

RESULTS

S-100 Protein Analysis

All individuals in the control group showed a nonde-tectable serum level of S-100 protein. Among the head-injured patients, the protein was detectable in 108 (39%)and nondetectable in 170 (61%). High serum levels weresignificantly correlated to lower GCS scores (regressionanalysis, r2 5 0.21, p 5 0.0001). Mean serum level wassignificantly (p 5 0.0001, ANOVA analysis) increasedin patients with severe injuries compared to those with

mild and moderate injuries (Table 1). All patients suf-fering severe head injury had detectable serum S-100 lev-els (mean, 3.6 m g/L; range, 1.2–12.5 m g/L) within 24 hpostinjury. Serum protein S-100 was detected in 75% ofmoderate head injury (mean, 0.7 m g/L; range, 0.2–2.2m g/L) and in 35% in mild head injury (mean, 0.6 m g/L;range, 0.2–6.2 m g/L).

Computed Tomography and Magnetic Resonance Imaging

Forty (14%) of the 278 patients demonstrated a patho-logic CT scan. Fifteen (5%) had isolated skull fracturewithout intracranial pathology, and 25 (9%) showed in-tracranial lesions (main pathology: brain contusion n 513, subdural hematoma n 5 6, epidural hematoma n 52, traumatic subarachnoid hemorrhage [tSAH] n 5 2, andbrain edema n 5 2). Among the 45 mildly injured pa-tients who underwent MRI, brain contusion was detectedin five (11%).

S-100 Protein Related to Neuroradiological Findings

Twenty-three (92%) of the 25 patients with intracra-nial lesions demonstrated detectable serum level of S-100protein, with a mean value of 2.2 (range, 0.2–12.5) m g/L.The sensitivity of S-100 measurements for detection ofintracranial pathology was 92%, and the specificity was66%. The positive predictive value of the test was 0.23,and the negative predictive value 0.99. Figure 1 showsthat the proportion of S-100–positive patients increasedwith more pronounced radiological evidence of braindamage. The proportion of S-100 positive patients wassignificantly (p 5 0.0001, x 2 test) increased in patientswith intracranial pathology (Table 2).

Figure 2 shows that mean S-100 level increased withmore pronounced neuroradiological findings. Patientswith severe head injury demonstrated significantly higherserum S-100 levels (p , 0.01, ANOVA) compared to pa-tients with moderate or mild head injury without in-tracranial lesions (groups mild A and mild B). Two ofthe 25 (8%) patients with intracranial pathology revealedby CT scan showed nondetectable serum levels. CT re-vealed small frontal contusions in the first patient (GCS14) and a minimal tSAH in the other (GCS 15).

DISCUSSION

The severity of TBI is difficult to assess, because spe-cific measures of the presence and severity of such in-juries have been unavailable. Therefore, there is a major

PROTEIN S-100 IN HEAD INJURY

643

need for a biological marker of the severity of TBI. Thisis especially true for mild injury in the medical-legal con-text in which it may be desirable or necessary to provethat disability or neuropsychological impairment after atraumatic event is really due to the head injury, in con-trast to stress disorder, systemic injury, or other causes.

The present report compares serum levels of S-100protein in patients with various severity grades of headinjury with healthy controls. We detected S-100 proteinin the serum of 39% of 278 head-injured patients. All pa-tients with severe head injury demonstrated increasedserum levels (1.2–12.5 m g/L). High serum levels weresignificantly correlated to both lower GCS scores and in-tracranial pathology revealed by neuroimaging. Unde-tectable serum levels predicted normal intracranial find-ings on CT.

In this report, head injuries were classified as eithermild, moderate, or severe. In addition, mild head injurieswere divided into four subgroups. Mild group A includedpatients with normal neuroradiological findings, whilegroups B, C, and D contained patients with various neu-roradiological abnormalities. A step-by-step increase inmean S-100 level was observed, with more pronouncedradiological evidence of brain damage. Hence, S-100measurements may add useful information on the sever-ity of brain damage in mild head injury.

S-100 Protein and Outcome After Head Injury

Woertgren et al. (1997) detected increased S-100serum levels in 83% of patients suffering severe head in-jury. Patients with bad outcome had significantly higher

ROMNER ET AL.

644

FIG. 1. Bar chart showing the proportion of S-100–positive patients. Total, all patients (n 5 278); mild A, patients with mildhead injury (GCS score 14–15) and normal neuroradiological findings (n 5 228); mild B, mild head injury and CT-verified skullfracture (n 5 14); mild C, mild head injury, normal CT scan, but cerebral contusions revealed by MRI (n 5 5); mild D, mild headinjury with CT-verified intracranial pathology (n 5 12); moderate, patients with moderate head injury (GCS score 9–13; n 5 16);and severe, patients with severe head injury (GCS score 3–8; n 5 8).

TABLE 2. RELA TION BETW EEN RESULTS O F S-100 PRO TEIN SERUM

MEA SU REM ENTS AN D FINDIN G S O N CO M PU TED TOM O G RAPH Y

S-100–positive group

Positive Negative Total

CTIntracranial pathology 23 (92%) 2 (8%) 25Normal findings 85 (34%) 168 (66%) 253

Total 108 (39%) 170 (61%) 278

CT, computed tomography.

protein S-100 concentrations compared to patients withgood outcome (mean, 4.9 versus 1.7 m g/L). Recently,Raabe et al. (1999) showed that elevated peak serum S-100 levels have a high predictive value for unfavorableoutcome after severe head injury. Studying 84 patientswith severe head injury, they found low levels in almostall patients who had a good outcome but increased lev-els in 57% of those who had not. Furthermore, all pa-tients who had a S-100 level higher than 3.8 m g/L died.The authors also described the finding of secondary ele-vations of S-100 in approximately half of the patientswho had an unfavorable outcome. Contrary to their find-ings, we observed S-100 serum levels up to 6.2 m g/L insurviving patients with moderate head injury and CT-ver-ified cerebral contusions.

In a detailed analysis of 50 MHI patients with normalCT scans, we showed that determination of S-100 pro-tein levels in serum provides a valid measure of the pres-ence and severity of TBI (Ingebrigtsen et al., 1999). De-tectable serum levels of S-100 protein were related to

brain contusion revealed by MRI and impaired neu-ropsychological functioning.

S-100 Protein and Pathophysiology of Head Injury

In the present study, 92% of the patients with in-tracranial lesions demonstrated detectable serum levelsof S-100 protein, with a mean value of 2.2 m g/L. In thetwo patients with nondetectable serum level but in-tracranial pathology revealed by CT, the blood sampleswere collected 1 and 7 h after the injury, respectively.We have previously shown that S-100 serum levels de-cline rapidly after a mild head injury (Ingebrigtsen etal., 1999). Six hours after injury, the protein is unde-tectable in almost 40% of those with initially detectablelevels. Raabe et al. (1999) found that S-100, althoughhaving a half-life of 2 h, remained pathologically in-creased after severe head injury. Thus, contrary to thepattern with minor head injury, S-100 is not only re-

PROTEIN S-100 IN HEAD INJURY

645

FIG. 2. Means and standard error bars in 108 head-injured patients with detectable S-100 serum levels. The dotted line de-

picts the detection limit of the analysis kit (0.2 m g/L). Total, all patients (n 5 108); mild A, patients with mild head injury(GCS score 14–15) and normal neuroradiological findings (n 5 69); mild B, mild head injury and CT-verified skull fracture,but normal intracranial findings (n 5 6); mild C, mild head injury, normal CT scan, but cerebral contusions revealed by MRI(n 5 3); mild D, mild head injury with CT-verified intracranial pathology (n 5 10); moderate, patients with moderate head in-jury (GCS score 9–13; n 5 12); and severe, patients with severe head injury (GCS score 3–8; n 5 8). Among patients withmild head injury, we observed higher serum levels of protein S-100 when CT or MRI revealed intracranial pathology, butthese differences were not statistically significant. Patients with severe head injury demonstrated significantly higher serum S-

100 levels (p , 0.01, ANOVA) compared to patients with moderate or mild head injury without intracranial lesions (groupsmild A and mild B).

leased at the time of primary injury but also during thefollowing days.

Little is known regarding the mechanism by which S-100 passes through the blood-brain barrier and enters theblood. The molecular weight is rather high (Mr 5 22,000)and may limit passage through the blood-brain barrier.Persisting elevated S-100 values may reflect either an on-going pathophysiolo gical cascade that leads to secondarybrain damage or the egress of S-100 as a result of necro-sis or apoptosis of previously damaged cells.

When Moore (1965) discovered the S-100 protein, hepresumed that it was located only in brain tissue. Subse-quent studies demonstrated its presence in the body indifferent forms, and the beta form predominates in thebrain (Haimoto et al., 1987; Jensen et al., 1985; Zimmeret al., 1995). Initially, S-100 was considered unique tothe glial and Schwann cells of the nervous system. It isnow clear that S-100 is present in other tissues also,mainly in adipocytes and chondrocytes. The concentra-tions in these cells, however, are very low (100–200ng/mg of protein), compared with concentrations in glialand Schwann cells (3,500 ng/mg of protein (Haimoto etal., 1987). Therefore, organs other than the brain are un-likely to be the source of the increased serum level of S-100 protein in head-injured patients.

It would be important to determine whether only spe-cific types of head injury pathophysiolo gy are responsi-ble for a protein S-100 increase. For example, S-100 pro-tein increase may specifically denote diffuse axonalinjury rather than agonist-induced neuroexcitation injuryor ischemic brain injury. Therefore, there is a need forbasic laboratory studies with this protein to determine itsimmunolocalization in models of head injury, itspredilection for white matter versus gray matter, andwhether it is generated more by diffuse axonal injury orischemic events.

Clinical Value of S-100 Protein in theManagement of Head Injury

The management of patients with mild head injury isusually focused on the risk of development of an acuteintracranial hematoma. In different studies, this life-threatening complication occurs in 0.7–4.0% of the pa-tients (Culotta et al., 1996; Gómez et al., 1996; Hsianget al., 1997; Stein, 1996). Today, routine early CT scanin usually recommended for early detection of suchhematomas (Stein, 1996). The present study indicates thatnondetectable S-100 protein levels in serum predict anormal CT scan if the blood sample is collected withinthe first hours after head injury. If this observation is re-producible in a larger number of patients, determinationof S-100 protein serum levels may replace routine CT

scans. If so, CT scanning would be necessary only in pa-tients with detectable serum levels.

Protein S-100 may also be useful for monitoring theefficacy of new therapies in both clinical and experi-mental settings to determine whether they influence theduration of increased protein S-100 levels and time nec-essary to achieve a decrease to normal levels.

CONCLUSION

S-100 protein is established as a serum marker of trau-matic brain injury. An undetectable serum level of S-100protein predicts normal intracranial findings on CT scan.Determination of S-100 protein in serum may be used toselect patients for scanning in situations where CT is inshort supply or where patients need to be sent long dis-tances to obtain a CT.

ACKNOWLEDGMENTS

We thank Drs. Eva A. Jacobsen and RagnhildSagsveen for help in reviewing the neuroradiological ex-aminations. The study was supported by The LærdalFoundation for Acute Medicine (Grant number 1629) andThe Skane County Council’s Research and DevelopmentFoundation. Sangtec Medical AB supplied the kits for S-100 protein analysis.

REFERENCES

BLOMQUIST, S., JOHNSSON, P., LÜHRS, C., et al. (1997).The appearance of S-100 protein in serum during and im-mediately after cardiopulmonar y bypass surgery: a possiblemarker for cerebral injury. J. Cardiothorac. Vasc. Anesth. 11,699–703.

CULOTTA, V.P., SEMENTILLI, M.E., GEROLD, K., andWATTS, C.C. (1996). Clinicopatholog ical heterogenity inthe classification of mild head injury. Neurosurgery 38,245–250.

GANDY, S.E., SNOW, R.B., ZIMMERMAN, R.D., andDECK, M.D.F. (1984). Cranial nuclear magnetic resonanceimaging in head trauma. Ann. Neurol. 16, 254–257.

GOMEZ, P.A., LABATO, R.D., ORTEGA, J.M., and DE LACRUZ, J. (1996). Mild head injury: differences in prognosisamong patients with a Glasgow Coma Scale score of 13 to15 and analysis of factors associated with abnorm al CT find-ings. Br. J. Neurosurg. 10, 453–460.

HAIMOTO, H., HOSODA, S., and KATO, K. (1987). Differ-ential distribution of immunoreactiv e S-100- a and S-100- bproteins in normal nonnervous human tissues. Lab. Invest.57, 489–498.

ROMNER ET AL.

646

HAN, J.S., KAUFMAN, B., ALFIDI, R.J., et al. (1984). Headtrauma evaluated by magnetic resonance and computed to-mography: a comparison. Radiology 150, 71–77.

HSIANG, J.N.K., YEUNG, T., YU, A.L.M., and POON, W.S.(1997). High-risk mild head injury. J. Neurosurg. 87,234–238.

INGEBRIGTSEN, T., ROMNER, B., KONGSTAD, P., andLANGBAKK, B. (1995). Increased serum concentrations ofprotein S-100 after minor head injury: a biochemical serummarker with prognostic value? J. Neurol. Neurosurg. Psy-chiatry 59, 103–104.

INGEBRIGTSEN, T., ROMNER, B., and TRUMPY, J.H.(1997). Management of minor head injury: the value of earlycomputed tomography and serum protein S-100 measure-ments. J. Clin. Neurosci. 4, 29–33.

INGEBRIGTSEN, T., WATERLOO, K., JACOBSEN, E.A.,LANGBAKK, B., and ROMNER, B. (1999). Traumatic braindamage in minor head injury: relation of serum S-100 pro-tein measurem ents to magnetic resonance imaging and neu-robehavioral outcome. Neurosurgery 45, 468–476.

JENKINS, A., TEASDALE, G., HADLEY, M.D.M., MAC-PHERSON, P., and ROWAN, J.O. (1986). Brain lesions de-tected by magnetic resonance resonance imaging in mild andsevere head injuries. Lancet 2, 445–446.

JENSEN, R., MARSHAK, D.R., ANDERSON, C., LUKAS,T.J., and WATTERSON, D.M. (1985). Characterization ofhuman brain S100 protein fraction: amino acid sequence ofS100 b . J. Neurochem. 45, 700–705.

LEVIN, H.S., AMPARO, E., EISENBERG, H.M., et al. (1987).Magnetic resonance imaging and computerized tomographyin relation to neurobehavioral sequelae of mild and moder-ate head injuries. J. Neurosurg. 66, 706–713.

MILD TRAUMATIC BRAIN INJURY COMMITTEE OFTHE HEAD INJURY INTERDISCIPLINARY SPECIALINTEREST GROUP OF THE AMERICAN CONGRESS OFREHABILITATION MEDICINE. (1993). Definition of mildhead injury. J. Head Trauma Rehabil. 8, 83–84.

MOORE, B.W. (1965). A soluble protein characteristic of thenervous system. Biochem. Biophys. Res. Commun. 19,739–744.

NYGAARD, Ø., LANGBAKK, B., and ROMNER, B. (1997).Age- and sex-related changes of S-100 protein concentrationsin cerebrospinal fluid and serum in patients with no previoushistory of neurological disorder. Clin. Chem. 43, 541–543.

RAABE, A., GROLMS, C., SORGE, O., ZIMMERMANN, M.,and SEIFERT, V. (1999). Serum S-100B protein in severehead injury. Neurosurgery 45, 477–483.

SHACKFORD, S.R., WALD, S.L., and ROSS, S.E. (1992). Theclinical utility of computed tomographic scanning and neu-rological examination in the management of patients with mi-nor head injuries. J. Trauma 33, 385–394.

STEIN, S.C. (1996). Management of minor closed head injury.Neurosurg. Q. 6, 108–115.

STEIN, S.C., and SPETELL, C. (1995). The head injury sever-ity scale (HISS): a practical classification of closed-head in-jury. Brain Inj. 9, 437–444.

USUI, A., KATO, K., ABE, T., MURASE, M., TANAKA, M.,and TAKEUTCHI, E. (1989). S-100a0 protein in blood andurine during open-heart surgery. Clin. Chem. 35, 1942–1944.

WATERLOO, K., INGEBRIGTSEN, T., and ROMNER, B.(1997). Neuropsycholog ical function in patients with in-creased serum levels of protein S-100 after minor head in-jury. Acta Neurochir. (Wien) 139, 26–32.

WOERTGREN, C., ROTHOERL, R.R., HOLZSCHUH, M.,METZ, C., and BRAWANSKI, A. (1997). Comparison ofserial S-100 and NSE serum measurement after severe headinjury. Acta Neurochir. (Wien) 139, 1161–1165.

ZIMMER, D.B., CORNWALL, E.H., LANDAR, A., andSONG, W. (1995). The S100 protein family: history, func-tion, and expression. Brain Res. Bull. 37, 417–429.

Address reprint requests to:Bertil Romner, M.D., Ph.D.

Department of NeurosurgeryUniversity Hospital, Lund

S-221 85 Lund, Sweden

E-mail: bertil.romner@neurokir.lu.se

PROTEIN S-100 IN HEAD INJURY

647