c 2010 wolters kluwer health lippincott williams & … low serum levels of s100b predict normal...
TRANSCRIPT
J Head Trauma RehabilVol. 25, No. 4, pp. 228–240Copyright c© 2010 Wolters Kluwer Health | Lippincott Williams & Wilkins
Can Low Serum Levels of S100BPredict Normal CT Findings AfterMinor Head Injury in Adults?:An Evidence-Based Review andMeta-Analysis
Johan Unden, MD, PhD; Bertil Romner, MD, PhD
Objective: To determine whether low levels of S100B in serum can predict normal computed tomography (CT)findings after minor head injury (MHI) in adults. Participants: Not applicable. Design: Systematic evidence-basedreview of the peer-reviewed literature with meta-analytical interpretation. Primary Measures: Not applicable. Results:We identified 12 eligible articles that specifically studied adult MHI patients with S100B and cranial CT scans in theacute phase after injury, comprising a total of 2466 separate patients. Individual negative predictive values of 90%to 100% were found for the ability of a negative (under cutoff) S100B level to predict a normal CT scan. A total of6 patients included in the studies had low S100B levels and positive CT scans (0.26%) and only 1 of these patients(0.04%) had a clinically relevant CT finding. The pooled negative predictive value for all studies was more than 99%(95% CI 98%–100%), with an average prevalence for any CT finding at 8%. The studies are consistently classed aslevel 2 and level 3 grades of evidence, suggesting a grade B recommendation. Conclusion: Low serum S100B levelsaccurately predict normal CT findings after MHI in adults. S100B sampling should be considered in MHI patientswith no focal neurological deficit, an absence of significant extracerebral injury, should be taken within 3 hours ofinjury, and the cutoff for omitting CT set at less than 0.10 μg/L. Care givers should also be aware of other clinicalfactors predictive of intracranial complications after MHI. Keywords: evidence, guidelines, head injury, management,meta-analysis, MHI, mild, minor, TBI, S100/S-100/S100B
TRAUMATIC HEAD INJURY is a significant causeof mortality and morbidity in adults1 and is the
leading cause of death in childhood.2 Minor head injury(MHI) represents up to 95% of head injuries3 and man-agement today involves computed tomography (CT)or in-hospital observation, although neither of theseoptions are ideal. In-hospital observation is not cost-effective4,5 and CT is sometimes impractical, not alwaysavailable, involves potentially harmful ionizing radia-tion, and is also relatively costly.6,7 Furthermore, onlya small proportion of patients with MHI have intracra-nial injuries (ICI) and even fewer require neurosurgical
Author Affiliations: Department of Anaesthesia and Intensive Care,Halmstad Regional Hospital, Halmstad, Sweden (Dr Unden); Departmentof Neurosurgery, Rigshospitalet, Copenhagen, Denmark (Dr Romner).
JU has received funding from the following noncommercial sources:Vetenskapliga Radet/Landstinget i Halland, Region Skane and SodraSjukvardsregionen.
Corresponding Author: Johan Unden, MD, PhD, Department of Anaesthesiaand Intensive Care, Halmstad Regional Hospital, 30185 Halmstad, Sweden([email protected]).
intervention.8,9 Several rules have been published aimedat identifying patients with higher risks for CT find-ings and/or neurosurgical intervention. These are basedon patient history and clinical examination such as theCanadian CT Head Rule, the New Orleans Criteria, CTin Head Injury, the Scandinavian Neurotrauma Com-mittee guidelines, and the National Institute for Clin-ical Excellence guidelines.9–13 These guidelines could,in theory, safely identify patients with intracranial com-plications after MHI and reduce CT usage by 12% to46% with very high sensitivities for ICI.14 However,in a recent evaluation, approximately 10% of physi-cians reported that Canadian CT Head Rule and NewOrleans Criteria guidelines were uncomfortable to ap-ply and approximately 5% of physicians misinterpretedthe guidelines and did not order a CT when one wasrequired.15 Furthermore, 30% to 50% of MHI patientsare intoxicated,16 which may confound several of theclinical predictors included in these decision rules.
Over the past 15 years, levels of protein S100B in bio-logical fluids have been shown to be correlated with the
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
228
Can Low Serum Levels of S100B Predict Normal CT Findings? 229
presence and severity of neurological disorders. S100Bwas originally thought to be specific to astroglia inthe central nervous system, but further studies haveshown the protein in neurons17 and in extracerebraltissues such as adipocytes, chondrocytes, lymphocytes,and bone marrow cells.18 The biological function ofS100B is still somewhat unclear, but it seems to haveboth toxic/degenerative and trophic/reparative roles de-pending on the concentration of the protein.19 S100B isfound in low, but measurable, levels in healthy individu-als, rises rapidly in blood after head injury and has a shorthalf-life of about 30 to 90 minutes.20,21 The possibilityof using serum S100B in patients with MHI was first re-ported in 1995.22 It was first thought that S100B releasewas a biomarker of subtle brain damage after MHI, al-though data suggest that an equally relevant mechanismmay involve the release of extracellular S100B througha disrupted blood-brain barrier, without necessarily in-volving actual cellular damage.19,23,24 Since the origi-nal studies from the mid-1990s, several studies from dif-ferent research groups have explored the ability of thisbiomarker to aid in decision making in the initial phasesof MHI management. These efforts concentrate on thehigh sensitivity of S100B for CT-evident injury, ratherthan specificity, as several studies have shown clear ex-tracerebral sources of the protein.25,26
This report will summarize and analyze the evidencecurrently available to attempt to answer the followingkey clinical question: Can low serum levels of S100Bpredict normal CT findings after minor head injury inadults?
METHODS
Definition of minor head injury
There exist several different definitions of MHI, butall have the common goal to classify patients accordingto the risk of developing acute intracranial complica-tions, such as intracranial hemorrhage and brain con-tusions. Most definitions of MHI include a short pe-riod of unconsciousness or amnesia and grading accord-ing to Glasgow Coma Scale (GCS). In 1995, Stein andSpettell presented their definition based on 25 000 pa-tients where patients with MHI are graded as GCS 14to 15.27 Patients with a GCS score of 13 are sometimesincluded in the MHI group despite the increased in-cidence of intracranial lesions in this group.28 In con-trast, including only patients with a GCS score of 15results in a patient group with a very low risk of clin-ically important complications.29 Mild traumatic braininjury (mTBI) is a term also used in the literature, essen-tially referring to an injury to the brain itself, rather thanthe process and presentation of the injury. This defini-tion is, however, problematic in the sense that it impliesa diagnosis of brain injury without the actual diagnostic
process being completed. Without elaborating furtheron nomenclature issues, MHI will be defined as followsfor the purpose of this review: History of nonpenetrat-ing head injury, GCS score 13 to 15 at admission, lossof consciousness (LOC) or amnesia. This wide defini-tion includes typical aspects from different MHI andmTBI definitions while still focusing on the process andpresentation of the head trauma, rather than the finaldiagnosis after neuroradiology or hospital stay. Studiesof patients presenting with GCS scores of 12 or less werenot included in the analysis.
Index test
The analysis of S100B in serum has been achievedthrough several different techniques, including im-munoradiometric assays, immunoluminometric assays,enzyme-linked immunosorbent assays, and electro-chemiluminescence immunoassays. These are avail-able from several commercial sources and differ inperformance.19 For the purpose of simplicity in this re-port, no distinction will be made between different as-says despite the fact that discrepancies in analytical per-formance may be of importance.30
Reference test
CT is not 100% sensitive for intracranial complica-tions after MHI.31–33 However, cranial CT is widely ac-cepted as the gold standard in detection of intracraniallesions after MHI and evidence shows that patients witha normal CT scan after MHI have a minimal risk of de-veloping an intracranial lesion.8 Cranial CT will there-fore be considered as the reference test in this report.
Search strategy and identification of relevant studies
To adequately answer the key question, we performeda Medline search for studies between 1983 (when CTbecame clinically available) to the present day using ap-propriate combinations of MeSH terms and key words;head injury, TBI, mTBI, MHI, minor, mild, mini-mal, serum, biomarkers, S-100, S100, S-100B, S100B, S-100BB, S100BB, computed tomography, CT, CCT, andmanagement. A less comprehensive TripDataBase andClinical Queries search using these keywords was alsoconducted. We performed this relatively wide search toinclude the maximum number of relevant patients. Thesearch was complemented by examining the bibliogra-phies of identified relevant studies.
Study eligibility
Studies containing adult patients with nonpenetrat-ing head injury with an admission/initial GCS scoreof 13 or more, S100B levels in serum and cranial CTwithin 24 hours of injury and possibilities for extraction
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
www.headtraumarehab.com
230 JOURNAL OF HEAD TRAUMA REHABILITATION/JULY–AUGUST 2010
of relevant data (sensitivities, specificities, positive pre-dictive values [PPV], negative predictive values [NPV],and prevalence) for the relevant patient group were in-cluded. Studies concerning children were excluded.
Data extraction
The eligible studies were examined and relevant datarecorded including; first author, year of publication,study design, patient group and inclusion criteria, char-acteristics of the index test including cutoff used, rele-vant results with respect to the key question includingpredictive statistics, and study limitations. If certain keyfactors or data were missing from the studies, authorswere contacted for clarification. In the case of multiplestudies from the same research group, authors were alsocontacted to ensure unique patients. Because a cutoff of0.10 μg/L has independently been reported from differ-ent research groups,21,34 results in relation to this levelwere extracted, if possible, to attempt an interpretationof data using the same cutoff.
Quality assessment
Studies were graded and assigned a quality rating withrespect to the key question according to the Centre ofEvidence Based Medicine (CEBM) criteria.35 Studieswere graded from 1 (strongest evidence, for instancereports of clinical decision rules and high quality val-idating study) to level 5 (weakest evidence, often expertopinion). Following this, the 14 Quality Assessment ofDiagnostic Accuracy Studies (QUADAS) criteria36 wereapplied. Studies receive scores from 1 (lowest quality) to14 (highest quality) based on 14 separate criteria relevantto diagnostic studies accounting for bias (items 3–7, 10–12), variability (items 1–2), and reporting (items 8–9, 13);1 point was given for each criteria satisfied. Studies with3 or more missing points (ie, 11 points or less in total)were downgraded one level. Studies therefore receiveda final evidence level accounting for both criteria. Thisapproach was used to account for classical features ofstudies (CEBM) including specific details related to di-agnostic studies (QUADAS). The final recommendation(A to D) is based on CEBM criteria; grade A referringto consistent class 1 studies, grade B for consistent level2 or 3 studies (or extrapolations from level 1 studies),grade C for level 4 studies (or extrapolations from level2 or 3 studies), and grade D for level 5 studies.
Data presentation and analysis
Studies are briefly presented in evidentiary tables.Data are presented in table form with correspondingnumber of patients with true positives (TPs), false pos-itives (FPs), false negatives (FNs), and true negatives(TNs) for each study along with relevant comments con-cerning FN patients. We explored heterogeneity using
a using a Chi-squared test. Because of heterogeneity,weighted pooled sensitivity and specificity were calcu-lated with a random effects model. We calculated likeli-hood ratios and predictive values from the pooled sen-sitivities and specificities derived from the random ef-fects model. This approach also eliminates differencesin prevalence from the studies and allows for calculationof predictive values corresponding to typical prevalencelevels for this patient group.
RESULTS
A total of 272 abstracts were examined independentlyby the authors and 49 full manuscripts were chosen andstudied in detail. Twelve studies were found eligible forthe current review and data analysis,21,34,37–46 consistingof 2466 separate patients. These studies are presented inthe evidentiary table, Table 1. The number of patientsincluded in the cohorts range from 50 to 1309 patients(average 206 patients). Time from injury to S100B sam-pling ranged from less than 3 hours to less than 24 hoursaccording to inclusion criteria (with all but one studysampling patients within 12 hours). The settings for allstudies were emergency departments in the followingcountries; Sweden, Norway, Germany, France, Slovenia,Brazil, and the United States. In 6 of the studies, dataconcerning a cutoff of 0.10μg/L could be analyzed, insome instances after contact with authors.
Table 2 presents studies with accompanying TP, FP,FN, and TN results, also including pathological radio-logical data from patients with S100B under the cutofflevel.
The 12 eligible studies show similar results with highindividual sensitivities (75%–100%). More importantly,they show very high individual NPVs (90%–100%) forthe ability of a negative (under cutoff level) S100B levelto predict a normal CT scan. Six FNs (low S100B, butpositive CT scans) were found in the total patient groupof 2264 patients (0.26%). These diagnoses were 2 smalltraumatic subarachnoid hemorrhages, 2 skull fractures,1 patient with a small cerebral contusion, and 1 patientwith an acute subdural hematoma. Only the latter pa-tient needed surgical intervention (1 patient from thetotal of 2264, 0.04%), 17 days after the trauma, fromneurological deterioration.
Meta-analysis
Sensitivities were only borderline homogenous(Q = 19, degrees of freedom = 11, P = 0.054) but speci-ficities were clearly heterogeneous (Q = 168, P < .001).Considering only those studies in which a cutoff of 0.10μg/L could be evaluated did not eliminate heterogeneity(Q = 15, degrees of freedom 7, P = .042 for sensitivity andQ = 27, degrees of freedom 7, P < .001 for specificity).Because of this heterogeneity, a random effects model
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Can Low Serum Levels of S100B Predict Normal CT Findings? 231
TAB
LE1
Evi
den
tiary
tabl
esh
owin
gel
igib
lest
udie
s21,3
4,3
7–46
wit
hre
leva
nt
info
rmati
on
Pa
tie
nt
Le
ve
lo
fE
vid
en
ce
gro
up
an
dIn
de
xR
ele
va
nt
Lim
ita
tio
ns
ev
ide
nce
,Q
UA
DA
Sle
ve
la
fte
rS
tud
y,
ye
ar
De
sig
nin
clu
sio
ncri
teri
ate
st,
cu
toff
S1
00
Bre
su
lts
an
dco
mm
en
tsC
EB
Msco
red
ow
ng
rad
ing
Inge
brig
tsen
etal
38P
rosp
ectiv
eco
hort
182
patie
nts
with
head
inju
ry,
≤10
min
LOC
,G
CS
scor
e13
–15,
nofo
cal
neur
olog
y,ag
e15
–80
y,ad
mis
sion
with
in12
hof
inju
ry,C
Tw
ithin
24h
ofin
jury
San
gtec
100®
,S
angt
ecM
edic
al,
Bro
mm
a,S
wed
en,
0.20
μg/
L
Sen
s0.
90,s
pec
0.65
,P
PV
0.13
,NP
V0.
99fo
rC
Tfin
ding
s2
patie
nts
need
edne
uros
urge
ry,b
oth
had
elev
ated
S10
0B.1
patie
ntm
isse
d(m
inim
altS
AH
)with
S10
0B7
hpo
stin
jury
.
Sel
ectio
nbi
as2
12II
Nyg
ren
deB
ouss
ard
etal
43
Pro
spec
tive
coho
rt66
patie
nts
with
head
inju
ry,L
OC
<30
min
,am
nesi
a<
24h,
GC
Ssc
ore
14–1
5,C
Tw
ithin
24h,
MR
Iwith
in7
d
LIA
,San
gtec
S10
0,S
angt
ecM
edic
al,
Bro
mm
a,S
wed
en,
0.10
μg/
L
Sen
s1.
00,s
pec
0.47
,P
PV
0.11
,NP
V1.
00fo
rC
Tfin
ding
s
Sen
s80
%at
cuto
ffof
0.15
μg/
LS
elec
tion
bias
.S
mal
lnum
ber
ofpa
tient
s.S
100B
sam
plin
gup
to24
haf
ter
inju
ry.
211
III
Baz
aria
net
al44
Nes
ted
coho
rt86
adul
tsw
ithhe
adin
jury
<4
hof
inju
ry,L
OC
<30
min
,am
nesi
a<
24h
GC
Ssc
ore
14–1
5
LIA
ISO
N,S
angt
ec10
0,D
iaS
orin
,0.
10μ
g/L
Sen
s0.
75,s
pec
0.65
,P
PV
0.13
,NP
V0.
90fo
rC
Tfin
ding
s1
patie
ntm
isse
dw
itha
skul
lfra
ctur
eat
0.10
μg/
Lcu
toff
Nes
ted
coho
rtsu
gges
ting
sele
ctio
nbi
as.
Sm
alln
umbe
rof
patie
nts.
212
II
Bib
erth
aler
etal
21P
rosp
ectiv
eco
hort
1309
isol
ated
head
inju
ry,a
dults
,w
ithin
3h
ofin
jury
,GC
Ssc
ore
13–1
5,1
orm
ore
of10
clin
ical
risk
fact
ors
Ele
csys
S10
0,R
oche
Dia
gnos
tics,
Man
nhei
m,
Ger
man
y,0.
10μ
g/L
Sen
s0.
99,s
pec
0.30
,P
PV
0.10
,NP
V0.
997
for
CT
findi
ngs
AU
C=
0.80
(95%
CI0
.75–
0.84
)1
patie
ntm
isse
dw
ithsk
ullfi
ssur
eno
tre
quiri
ngtr
eatm
ent
Larg
eno
rmal
cont
rolg
roup
(n=
540)
.S
elec
tion
bias
.
212
II
(con
tinue
s)
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
www.headtraumarehab.com
232 JOURNAL OF HEAD TRAUMA REHABILITATION/JULY–AUGUST 2010
TAB
LE1
Evi
den
tiary
tabl
esh
owin
gel
igib
lest
udie
s21,3
4,3
7–46
wit
hre
leva
nt
info
rmati
on(C
ontin
ued)
Pa
tie
nt
Le
ve
lo
fE
vid
en
ce
gro
up
an
dIn
de
xR
ele
va
nt
Lim
ita
tio
ns
ev
ide
nce
,Q
UA
DA
Sle
ve
la
fte
rS
tud
y,
ye
ar
De
sig
nin
clu
sio
ncri
teri
ate
st,
cu
toff
S1
00
Bre
su
lts
an
dco
mm
en
tsC
EB
Msco
red
ow
ng
rad
ing
Bib
erth
aler
etal
41P
rosp
ectiv
eco
hort
104
patie
nts
with
isol
ated
head
inju
ryw
ithin
2h
ofin
jury
Pos
itive
CT
scan
resu
lt
LIA
ISO
N,S
angt
ecD
iagn
ostic
a,D
ietz
enba
ch,
Ger
man
y,0.
12ng
/mL
Sen
s1.
00,s
pec
0.46
,P
PV
0.13
,NP
V1.
00fo
rC
Tfin
ding
sA
UC
=0.
79(9
5%C
I0.
7–0.
89)
ICIp
reva
lenc
eof
23%
,sug
gest
ing
sele
ctio
nbi
as
211
III
Mul
ler
etal
,200
7P
rosp
ectiv
eco
hort
226
adul
tsw
ithhe
adin
jury
,<
12h
ofin
jury
,LO
C/a
mne
sia,
GC
Ssc
ore
13–1
5,no
foca
lne
urol
ogic
alde
ficit,
nom
ultip
lein
jurie
s,no
rena
l/liv
erdi
seas
e
LIA
ISO
N,D
iaS
orin
AB
,Bro
mm
a,S
wed
en,
0.10
μg/
L
Sen
s0.
95,s
pec
0.31
,P
PV
0.12
,NP
V0.
98fo
rC
Tfin
ding
sA
UC
0.73
(95%
CI0
.62–
0.84
).1
patie
ntm
isse
dw
ithsm
allb
rain
cont
usio
nno
tre
quiri
ngtr
eatm
ent
Sel
ectio
nbi
as2
12II
Mus
sack
etal
40P
rosp
ectiv
eco
hort
139
adul
tsw
ithhe
adin
jury
,<3
hof
inju
ry,G
CS
scor
e13
–15,
1of
5ris
ksy
mpt
oms
LIA
ISO
N,
Byk
-San
gtec
,D
iagn
ostic
a,G
erm
any,
0.21
μg/
L
Sen
s1.
0,sp
ec0.
5,P
PV
0.24
,NP
V1.
0fo
rC
Tfin
ding
sA
UC
0.86
(95%
CI0
.79–
0.94
)
Sel
ectio
nbi
as.N
SE
and
gluc
ose
not
sign
ifica
nt.
212
II
Pol
i-de-
Figu
eire
doet
al39
Pro
spec
tive
coho
rt,
valid
atio
nof
pred
e-te
rmin
edcu
toff
leve
l
50pa
tient
sw
ithhe
adin
jury
,<3
hof
inju
ry,G
CS
scor
e13
–15,
nofo
caln
euro
logi
cal
defic
its,1
of6
risk
sym
ptom
s
Ele
csys
2010
,R
oche
Dia
gnos
tics,
Man
nhei
m,
Ger
man
y,0.
10μ
g/L
Sen
s1.
0,sp
ec0.
2,P
PV
0.15
,NP
V1.
0fo
rC
Tfin
ding
sA
UC
0.82
(95%
CI
0.69
–0.9
6)
Sel
ectio
nbi
as.
Sm
alln
umbe
rof
patie
nts.
212
II
(con
tinue
s)
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Can Low Serum Levels of S100B Predict Normal CT Findings? 233
TAB
LE1
Evi
den
tiary
tabl
esh
owin
gel
igib
lest
udie
s21,3
4,3
7–46
wit
hre
leva
nt
info
rmati
on(C
ontin
ued)
Pa
tie
nt
Le
ve
lo
fE
vid
en
ce
gro
up
an
dIn
de
xR
ele
va
nt
Lim
ita
tio
ns
ev
ide
nce
,Q
UA
DA
Sle
ve
la
fte
rS
tud
y,
ye
ar
De
sig
nin
clu
sio
ncri
teri
ate
st,
cu
toff
S1
00
Bre
su
lts
an
dco
mm
en
tsC
EB
Msco
red
ow
ng
rad
ing
Inge
brig
tsen
etal
37P
rosp
ectiv
eco
hort
50pa
tient
sw
ithhe
adin
jury
,LO
C<
20m
in,G
CS
scor
e13
–15,
nofo
caln
euro
logi
cal
defic
its,n
oC
T-ev
iden
tin
jury
IRM
A,S
angt
ec,
Bro
mm
a,S
wed
en.
0.20
μg/
L
Sen
s1.
0,S
pec
0.77
,P
PV
0.21
,NP
V1.
0fo
rC
Tfin
ding
s
CT-
verifi
edin
jurie
sno
tin
clud
ed.
MR
Iper
form
edin
45pa
tient
s,se
vera
lpa
thol
ogic
alre
sults
,1of
thes
ew
ithS
100B
unde
rcu
toff
.Sel
ectio
nbi
as.S
mal
lnu
mbe
rof
patie
nts.
211
III
Bib
erth
aler
etal
42P
rosp
ectiv
eco
hort
52pa
tient
sw
ithhe
adin
jury
,GC
Ssc
ore
13–1
5,1
of6
risk
sym
ptom
s,no
foca
lne
urol
ogic
alde
ficits
LIA
,San
gtec
,B
yk-S
angt
ec,
Ger
man
y,0.
10μ
g/L
Sen
s1.
0,sp
ec0.
41,
PP
V0.
41,N
PV
1.0
for
CT
findi
ngs
ICIp
reva
lenc
eof
29%
,sug
gest
ing
sele
ctio
nbi
as.
Sm
alln
umbe
rof
patie
nts.
211
III
Bou
vier
etal
46P
rosp
ectiv
eco
hort
105
patie
nts
with
head
inju
ry,G
CS
scor
e13
–15
or1
of10
risk
fact
ors
Ele
csys
2010
,R
oche
Dia
gnos
tics,
Man
nhei
m,
Ger
man
y,0.
10μ
g/L
Sen
s1.
0,sp
ec0.
33,
PP
V0.
21,N
PV
1.0
for
CT
findi
ngs
AU
C0.
83(9
5%C
I0.
74–0
.89)
Art
icle
tran
slat
edfr
omFr
ench
.S
elec
tion
bias
.S
ens
and
NP
V10
0%ev
enat
cuto
ff0.
15μ
g/L.
212
II
Mor
ocho
vic
etal
45P
rosp
ectiv
eco
hort
97pa
tient
sw
ithhe
adin
jury
,GC
Ssc
ore
13–1
5w
ithor
with
out
risk
fact
ors
Ele
csys
,S10
0,R
oche
Dia
gnos
tics,
Man
nhei
m,
Ger
man
y,0.
10μ
g/L
Sen
s0.
88,s
pec
0.28
,P
PV
0.21
,NP
V0.
92fo
rC
Tfin
ding
s2
patie
nts
mis
sed,
1w
ithtS
AH
and
1w
ithaS
DH
ICIp
reva
lenc
e18
%.S
elec
tion
bias
.Sm
all
num
ber
ofpa
tient
s.aS
DH
did
not
need
surg
ery
until
afte
r17
d.
211
III
Abb
revi
atio
ns:
aSD
H,
acut
esu
bdur
alhe
mat
oma;
AU
C,
area
unde
rth
ecu
rve;
CE
BM
,C
entr
eof
Evi
denc
eB
ased
Med
icin
e;C
T,co
mpu
ted
tom
ogra
phy;
ICI,
intr
acra
nial
inju
ries;
LOC
,lo
ssof
cons
ciou
snes
s;N
SE
,ne
uron
-spe
cific
enol
ase;
MR
I,m
agne
ticre
sona
nce
imag
ing;
NP
V,ne
gativ
epr
edic
tive
valu
e;P
PV,
posi
tive
pred
ictiv
eva
lue;
QU
AD
AS
,Q
ualit
yA
sses
smen
tof
Dia
gnos
ticA
ccur
acy
Stu
dies
;sen
s,se
nsiti
vity
;spe
c,sp
ecifi
city
;tS
AH
,tra
umat
icsu
bara
chno
idhe
mor
rhag
e.
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
www.headtraumarehab.com
234 JOURNAL OF HEAD TRAUMA REHABILITATION/JULY–AUGUST 2010
TABLE 2 Table showing the 12 eligible studies with true positives (S100+, CT+), falsepositives (S100+, CT−), false negatives (S100−, CT+), true negatives (S100−, CT−)and brief information concerning false negatives
CT+ Patientswith S100
Study, year S100+, CT+ S100+, CT− S100−, CT+ S100−, CT− Total Below Cutoff
Ingebrigtsen et al38 9 60 1 112 182 tSAH, S100B 7 hafter injury, notreatment
Bazarian et al44 3 73 1 9 86 Skull fractureNygren de Boussard et al43 4 33 0 29 66Poli-de-Figueiredo et al39 6 35 0 9 50Muller et al, 2007 20 141 1 64 226 Small brain
contusion, nobrain injury ortreatment
Ingebrigtsen et al37 3 11 0 36 50Biberthaler et al21 92 855 1 361 1309 Skull fracture, no
brain injury ortreatment
Biberthaler et al41 24 43 0 37 104Biberthaler et al42 15 22 0 15 52Mussack et al40 19 60 0 60 139Bouvier et al46 16 60 0 29 105Morochovic et al45 15 57 2 23 97 tSAH and SDH
(83 y of age,GCS score 15,surgery after17 d)
Total 226 1450 6 784 2466
Abbreviations: CT, computed tomography; GCS, Glasgow Coma Scale; tSAH, traumatic subarachnoid hemorrhage; SDH, subduralhematoma.
was utilized. Figures 1 and 2 show graphical presenta-tions of sensitivities and specificities from the studies,and Figure 3 shows a summary receiver operating char-acteristic curve based on the same data. Table 3 presentsthe weighted pooled statistical data from the 12 stud-ies. The pooled sensitivity for all studies was 97% (95%CI 91%–99%) and the pooled specificity 40% (95% CI30%–51%). Considering the 6 studies where a cutoff of0.10 μg/L could be evaluated, sensitivities and specifici-ties were 96% (95% CI 85%–99%) and 30% (95% CI23%–38%), respectively.
The prevalence of intracranial findings after MHIhas been reported to be in the ranges of 1% to10%.8,9
Corresponding NPVs for prevalence levels of 1%, 5%,10%, and 20% are 100% (95% CI 100%–100%), 100%(95% CI 99%–100%), 99% (95% CI 97%–100%), and98% (95% CI 94%–99%), respectively. PPVs consider-ing prevalence levels of 1%, 5%, 10%, and 20% are 2%(95% CI 1%–2%), 8% (95% CI 7%–9%), 15% (95% CI13%–18%), and 29% (95% CI 25%–33%), respectively.The average prevalence from the included studies in this
article was 8%, giving a NPV of more than 99% (95%CI 98%–100%).
The studies are consistently classed as level 2 and level3 grades of evidence, which suggests a grade B recom-mendation according the CEBM criteria.
DISCUSSION
Limitations
The included studies use different assays for detec-tion of S100B in serum, which is a potential sourceof error. Different assays may report different levels ofS100B from the same sample from differences in per-formance characteristics.30 Four studies used the LIAI-SON assay from Sangtec (total number of patients 555),2 used the Elecsys assay from Roche (total number of pa-tients 1359), 2 used immunoluminometric assays fromSangtec (total number of patients 118), 1 used immuno-radiometric assays from Sangtec (total number of pa-tients 50), and the final study used a Sangtec 100 (to-tal number of patients 182). Hence, the Elecsys assay
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Can Low Serum Levels of S100B Predict Normal CT Findings? 235
Figure 1. Figure showing individual and pooled sensitivity ofS100B for computed tomography findings in 12 studies. Sizesof bold markers proportional to logarithms of number of pa-tients in the studies; horizontal lines represent 95% confidenceintervals computed according to Agresti and Coull.
accounts for 60% of the patients and the LIAISON ac-counts for 25%. Because the LIAISON assay reportshigher values of S100B than the Elecsys assay,30 it isunlikely that the pooled sensitivity would be decreasedif differences in assay performance were corrected. Ulti-mately, it is most correct to apply evidence from a cer-tain assay to only those patients measured with the sameassay. This approach is unpractical however, because dif-
Figure 2. Figure showing individual and pooled specificity ofS100B for computed tomography findings in 12 studies. Sizesof bold markers proportional to logarithms of number of pa-tients in the studies; horizontal lines represent 95% confidenceintervals computed according to Agresti and Coull.
Figure 3. sROC (summary receiver operating characteristic)curve from all 12 studies showing the relationship of sensitivityvs 1-specificity.
ferent assays will always exist and new assays with newtechnologies will replace older versions. This problem istherefore not unique to S100B and difficult to accountfor.
Because S100B is also produced by melanocytes, dif-ferent races could have different basal levels of S100B.This aspect has not been examined in these studies andshould be considered in future studies.
According to the QUADAS tool, the included pa-tients are judged relatively representative for the tar-get population for the test. Despite this, the prevalenceof intracranial complication differs between the stud-ies, generally being higher than values reported in theliterature.8,9 In particular, 2 of the studies had a preva-lence of CT findings over 20%, yet both these alsoshowed NPVs of 100%.41,42 These aspects, together withthe observed heterogeneity, are worrying and suggestbias but cannot be penetrated further in this report.However, consideration for this was achieved through
TABLE 3 Data showing pooled statisticsfor all patients from the 12 studies
Lower 95% Upper 95%CI CI
Sensitivity 97 91 99Specificity 40 30 51Positive 1.6 1.3 1.9
likelihood ratioNegative 0.080 0.025 0.25
likelihood ratioTheoretical 32%
computedtomographyreduction
Abbreviation: CI, confidence interval.
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
www.headtraumarehab.com
236 JOURNAL OF HEAD TRAUMA REHABILITATION/JULY–AUGUST 2010
the use of a random effects model for the statistical anal-ysis and calculating predictive values based on typicalprevalence values for this patient group.
Detection of CT findings versus surgical intervention
Studies concerning the early diagnostic managementof head injury usually report both CT findings andneurosurgical intervention (alternatively clinically im-portant brain injury) as endpoints.9–11,13 In this report,we focus only on CT findings. However, it is obvious,without statistical analysis, that the sensitivity of S100Bfor surgical intervention (or clinically important braininjury) would be even higher because only 1 patient(subdural hematoma) would be included in this group.Because the prevalence of surgical intervention and clini-cally important brain injury is very low after MHI (0.1%–1%),8,9 the NPV based on the data would be 100% withvery narrow CIs.
Can S100B be used in other severities of head injury?
The predictive performance of S100B will depend onwhich patients are considered for testing. This is relatedto the pretest probability (prevalence) of the relevant out-come; in this case, CT findings or clinically relevant in-tracranial complication. A lower pretest probability willresult in an increased NPV and a decreased PPV andvice versa. This also stresses that S100B can initially beapplied only to similar patient cohorts as those reportedin the studies examined within this report. ApplyingS100B to other patients will result in different posttestprobabilities, which may cause a diagnostic imbalancetoward false negatives (missed patients) or false positives(unnecessary CT scans). Oh et al found a sensitivity of97%, a specificity of 54%, an NPV of 88%, and a PPVof 84% for CT or magnetic resonance imaging findingsusing a cutoff of 0.105 μg/L in a cohort of 45 patientswith mild, moderate, and severe head injury.47 In fact,only 1 patient with radiological findings showed a S100Bbelow the reference limit, despite the high prevalenceof complications and the use of MRI as the referencetest.
The emergence of new decision rules for CT scan-ning has further complicated this issue. Ideally, S100Bcould be used in conjunction with these rules. Becauseof the very high sensitivity of S100B, the test should bemade before applying the decision rule when the pretestprobability is low. Applying S100B to patients selectedfor CT (via the decision rule) would lead to a lower nega-tive prediction because of a higher pretest probability. Itmust be remembered that, in contrast to the clinical deci-sion rules, S100B is an objective measurement and is notaffected by intoxication in MHI. Studies of CT predic-tion combining S100B with recently presented clinicaldecision rules are welcomed.
Is S100B influenced by alcohol intoxication?
Clinical parameters included in existing and proposedmanagement guidelines may be difficult to interpret inintoxicated patients, and some of these guidelines in-clude intoxication as a indicator for CT scanning.9,48 Inaddition, CT scanning or hospital observation may bepractically difficult in these patients. Because between30% and 50% of patients with MHI are intoxicated,49,50
this presents a practical management problem. Small el-evations of S100B after moderate alcohol intake in non–head-injured subjects has been reported,51 althoughS100B does not seem to be affected by alcohol intoxi-cation in head-injured patients.40,51,52 This is confirmedby unpublished data from our clinic where intoxicatedpatients show similar levels of S100B compared withsober patients.
What is the evidence for S100B use in children withhead injury?
The management of children with traumatic head in-jury is more complicated than in adults, because pa-tient history may be unreliable and physical exami-nation can be practically difficult. Also, CT scans ofchildren should be minimized because of potentialhealth risks.7 Despite recent impressive reports of pre-diction rules,53 a biomarker is welcomed in this patientcategory. Reference levels for healthy children are higherthan in adults.54–56 Berger and colleagues have publishedseveral studies concerning S100B levels in children afterTBI,57–60 showing promising results. These studies alsohighlight the diagnostic problem of inflicted TBI, an areawhere a reliable biomarker may have considerable clini-cal impact. Bechtel et al61 recently reported a study withS100B levels in serum after closed head injury in chil-dren younger than 18 years. They found that S100B wasa poor predictor of intracranial injury at a cutoff of 0.5μg/L with a PPV of only 20%, but found a high NPVof 90% at the same cutoff. They also investigated theimpact of bone fractures on S100B levels and, similar toresults in adult studies,25 found that these extracranialinjuries may be responsible for the poor specificity ofS100B in trauma. Castellani et al present a study of 109children with MHI in which they found a sensitivity andNPV of 100%, a specificity of 42%, and a PPV of 46%for CT findings using a cutoff of 0.16 μg/L derived fromanalysis of healthy children.62
Venous cannulation is not always performed afterMHI in children, especially in asymptomatic cases. Re-ports considering urine measurements of S100B in chil-dren showed initial promise.63,64 However, in a case-control pilot study, Pickering et al concluded that early(<12 hours after injury) urinary levels of S100B arenot useful in pediatric head injury.65 A recent study bySchultke et al confirmed these findings.66 Measurements
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Can Low Serum Levels of S100B Predict Normal CT Findings? 237
of capillary S100B would be a suitable option becausethis sampling method is often used for children in theemergency department. Unpublished results from an on-going study in Sweden show that capillary S100B is sig-nificantly correlated to, but not interchangeable with,venous samples. Therefore, separate reference and in-jury values must be studied before this method can beconsidered. Summarizing, although reports are gener-ally promising, the evidence for S100B in pediatric TBIis insufficient and must be expanded before it can beconsidered for clinical utility.
What is the evidence for other biomarkersin MHI management?
Because S100B is not specific to the brain, othermore specific biomarkers are needed to obtain a clin-ically useful specificity and PPV for brain injury afterMHI. Several other biomarkers have been studied inhead injury, including glial fibrillary acidic protein,67–69
brain- and heart- fatty acid-binding proteins,70 myelinbasic protein,71 neuron-specific enolase,72 and alpha II-spectrin breakdown products.73 Of these, glial fibrillaryacidic protein seems most interesting with a much betterclinical specificity than S100B.74 However, these mark-ers have not been sufficiently investigated in MHI andfurther studies are needed before any conclusions canbe made.
Will S100B miss important CT complications?Comparison with other clinical tests
S100B used in the proposed setting is not 100% sen-sitive. It is unreasonable to expect an actual 100% sen-sitivity or NPV from a diagnostic test. In this clinicalsituation, a very high sensitivity and even higher NPV isacceptable if combined with other diagnostic variables.It is important that clinicians are aware that neither de-cision rules nor recommendations are completely reli-able but merely based on the best available evidence.75
If S100B was to be used alone as a diagnostic tool af-ter MHI, without considering other clinical aspects, thiswill eventually lead to a false negative result (ie, a caseof ICI after MHI will be missed). Parallels can be madewith other diagnostic tests in potentially serious clinicalconditions.
Pulmonary embolism (PE) is a feared clinical con-dition with relatively high morbidity and mortality.Diagnostic problems with PE show several parallels withintracranial complications after MHI. The blood testD-dimer has been used for many years for the purposeof selecting patients who do not need radiological in-vestigation. Despite many years of experience with thetest and multiple efforts of refining assays and guide-lines, the sensitivity and NPV of D-dimer for PE has
recently been reported to be 95% (95% CI, 73.1%–99.7%) and 99% (95% CI, 96.2%–99.9%), respectively,with low specificity and PPV.76 This is inferior to the per-formance of S100B. The performance of D-dimer is evenpoorer when referring to deep vein thrombosis insteadof PE with sensitivities/NPVs of 88%/99%, 90%/96%,and 92%/84% in low-, moderate-, and high-prevalencegroups, respectively.77
Myoglobin, troponin I, and troponin T have largelyreplaced the rather unspecific creatine kinase MB asa heart-specific biomarker used in the diagnostics ofacute myocardial infarction. In a recent large multi-center study, Keller et al found sensitivities/NPVs foracute myocardial infarction of 61%/87% for myoglobin,91%/96% for troponin , and 73%/91% for troponin T.78
Epidural hematoma is a feared complication to MHI.Initially, before clinical manifestation, this conditionmay not be associated with any actual brain damage, butis rather an expanding extradural mass with a risk of fu-ture brain damage if left untreated. Hence, a biomarkerof brain damage may be insensitive of this condition.However, all epidural hematomas reported in the litera-ture have shown S100B levels over 0.10 μg/L, althoughthe values are often relatively low,34,79 which supportsthe idea that S100B is also a biomarker of blood-brainbarrier disruption23,24 as well as a marker of actual celldamage. Six patients from this report had pathologicalCT findings and a S100B level below cutoff. Only 1 ofthese patients required surgical treatment and would beclassified as clinically important brain injury accordingto the Canadian CT Head Rule decision rule.10 This pa-tient (an 83-year-old female with cardiopulmonary dis-ease) had an admission GCS score of 15 with risk fac-tors of headache and vomiting and would therefore bemissed by several of the existing clinical guidelines. Also,this patient was first surgically treated after 17 days post-trauma for neurological deterioration (presumably thena chronic subdural hematoma and hence likely to beeventually detected by clinical means).45
The half-life of S100B has been shown to be as shortas 25 minutes,20 although a recent study in patients withMHI has shown a half-life of 97 minutes.80 This impliesthat the timing of S100B sampling is important to avoidmissing patients with CT pathologies. If considered forclinical use, S100B should therefore be taken within 3hours of injury until more data are available.
Certain clinical risk factors have shown strong associ-ations with intracranial injury. In our opinion, patientswith special risk factors, particularly clinical signs of skullfracture, posttraumatic seizures, certain anticoagulationtherapies, and shunt-treated hydrocephalus should notundergo S100B sampling, but should receive a CT ac-cording to normal practice,12 because these factors havenot been properly addressed in the literature.
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
www.headtraumarehab.com
238 JOURNAL OF HEAD TRAUMA REHABILITATION/JULY–AUGUST 2010
Future aspects
Serum S100B levels show a high sensitivity and NPVfor intracranial complications after MHI. The speci-ficity is poor, mostly from contamination of extracere-bral S100B,25 although it is also possible that S100Bmay be more sensitive for certain trauma-induced le-sions than CT81 and elevated levels may, in some cases,be reflecting subtle brain damage. It is also possible thatextracranial trauma may release cytokines that can dis-rupt the blood-brain barrier and allow extracellular cere-bral S100B to enter the bloodstream. The developmentand validation of other, more specific brain biomarkers,should address these aspects. Specificity issues aside, thisobjective biochemical tool may reduce the frequency ofCT scanning after MHI by as much as 30%. Incorpo-rating methods of correction to account for extracere-bral contamination may further improve this figure.44
These are, however, only theoretical values of CT reduc-tion and the health economics of S100B implementa-tion remain to be elucidated. An interesting report fromRuan et al addresses this aspect and the authors concludethat S100B can be cost saving, depending on the clin-ical utility and setting.82 Preliminary data from an on-going study in a level III trauma hospital in Halmstad,Sweden, indicate that S100B has reduced CT scans after
MHI by approximately 15% and is cost effective. Dur-ing 3 years of clinical application, no patients have beenmissed by S100B, currently used as a clinical decisionrule in conjunction with the Scandinavian NeurotraumaCommittee guidelines. Several studies concerning theuse of brain biomarkers, including S100B, in pediatrichead injury are underway.
Serum levels of S100B are used in MHI managementin several European countries, including Sweden andGermany, in conjunction with existing guidelines, al-though none of these has been published or validated.
CONCLUSION
Low serum S100B levels accurately predict normal CTfindings after MHI in adults. The evidence in this reportsupports a grade B recommendation. S100B samplingshould be considered in MHI patients with no focalneurological deficit, an absence of significant extracere-bral injury, should be taken within 3 hours of injury andthe cutoff for omitting CT set at less than 0.10 μg/L.Approximately one third of CT scans may be omit-ted using this approach in the defined patient group,although care givers should be aware of other clini-cal factors predictive of intracranial complications afterMHI.
REFERENCES
1. Jennett B. Epidemiology of head injury. Arch Dis Child.1998;78(5):403–406.
2. Sharples PM, Storey A, Aynsley-Green, et al. Causes of fatal child-hood accidents involving head injury in northern region 1979-86.BMJ. 1990;301:1193–1197.
3. Cassidy JD, Carroll LJ, Peloso PM, et al. WHO CollaboratingCentre Task Force on Mild Traumatic Brain Injury. Incidence,risk factors and prevention of mild traumatic brain injury: resultsof the WHO Collaborating Centre Task Force on Mild TraumaticBrain Injury. J Rehabil Med. 2004(43) (suppl):28–60.
4. Ingebrigtsen T, Romner B. Routine early CT-scan is cost savingafter minor head injury. Acta Neurol Scand. 1996;93(2/3):207–210.
5. Norlund A, Marke LA, af Geijerstam JL, Oredsson S, Brit-ton M. OCTOPUS Study. Immediate computed tomographyor admission for observation after mild head injury: cost com-parison in randomised controlled trial. BMJ. 2006;333(7566):469.
6. Reinus WR, Wippold FJ II, Erickson KK. Practical selection cri-teria for noncontrast cranial computed tomography in patientswith head trauma. Ann Emerg Med. 1993;22:1148–1155.
7. Hall EJ, Brenner DJ. Cancer risks from diagnostic radiology. Br JRadiol. 2008;81(965):362–378. Review 1148–1155.
8. af Geijerstam JL, Oredsson S, Britton M. OCTOPUS StudyInvestigators. Medical outcome after immediate computed to-mography or admission for observation in patients with mildhead injury: randomised controlled trial. BMJ. 20062;333(7566):465.
9. Haydel MJ, Preston CA, Mills TJ, Luber S, Blaudeau E, DeBlieuxPM. Indications for computed tomography in patients with minorhead injury. N Engl J Med. 2000;343(2):100–105.
10. Stiell IG, Wells GA, Vandemheen K, et al. The CanadianCT Head Rule for patients with minor head injury. Lancet.2001;357(9266):1391–1396.
11. Smits M, Dippel DW, Steyerberg EW, et al. Predicting intracra-nial traumatic findings on computed tomography in patients withminor head injury: the CHIP prediction rule. Ann Intern Med.2007;146(6):397–405.
12. Ingebrigtsen T, Romner B, Kock-Jensen C. Scandinavian guide-lines for initial management of minimal, mild, and moderate headinjuries. The Scandinavian Neurotrauma Committee. J Trauma.2000;48(4):760–766.
13. Yates DW. The NICE head injury guidelines. Emerg Med J.2003;20(2):117.
14. Stein SC, Fabbri A, Servadei F, Glick HA. A critical comparison ofclinical decision instruments for computed tomographic scanningin mild closed traumatic brain injury in adolescents and adults.Ann Emerg Med. 2009;53(2):180–188.
15. Stiell IG, Clement CM, Rowe BH, et al. Comparison of the Cana-dian CT Head Rule and the New Orleans Criteria in patients withminor head injury. JAMA. 2005;,294(12):1511–1518.
16. Parry-Jones BL, Vaughan FL, Miles Cox W. Traumatic brain injuryand substance misuse: a systematic review of prevalence and out-comes research (1994-2004). Neuropsychol Rehabil. 2006;16(5):537–560.
17. Steiner J, Bernstein HG, Bielau H, et al. Evidence for a wide extra-astrocytic distribution of S100B in human brain. BMC Neurosci.20072;8:2.
18. Donato R. S100: a multigenic family of calcium-modulatedproteins of the EF-hand type with intracellular and extracel-lular functional roles. Int J Biochem Cell Biol. 2001;33(7):637–668.
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Can Low Serum Levels of S100B Predict Normal CT Findings? 239
19. Goncalves CA, Leite MC, Nardin P. Biological and methodolog-ical features of the measurement of S100B, a putative marker ofbrain injury [published online ahead of print April 18, 2008]. ClinBiochem. 2008;41(10/11):755–763. Review.
20. Jonsson H, Johnsson P, Hoglund P, Alling C, Blomquist S. Elim-ination of S100B and renal function after cardiac surgery. J Car-diothorac Vasc Anesth. 2000;14(6):698–701.
21. Biberthaler P, Linsenmeier U, Pfeifer KJ, et al. Serum S-100Bconcentration provides additional information for the indicationof computed tomography in patients after minor head injury: aprospective multicenter study. Shock. 2006;25(5):446–453.
22. Ingebrigtsen T, Romner B, Kongstad P, Langbakk B. Increasedserum concentrations of S-100 after minor head injury. A bio-chemical marker with prognostic value? J Neurol Neurosurg Psychi-atry. 1995;59:103–104.
23. Kleindienst A, Ross Bullock M. A critical analysis of the role of theneurotrophic protein S100B in acute brain injury. J Neurotrauma.2006;23(8):1185–1200.
24. Kanner AA, Marchi N, Fazio V, et al. Serum S100beta: a nonin-vasive marker of blood-brain barrier function and brain lesions.Cancer. 2003;1:97.
25. Unden J, Bellner J, Eneroth M, Alling C, Ingebrigtsen T, RomnerB. Raised serum S100B levels after acute bone fractures withoutcerebral injury. J Trauma. 2005;58(1):59–61.
26. Anderson RE, Hansson LO, Nilsson O, Dijlai-Merzoug R, Set-tergren G. High serum S100B levels for trauma patients withouthead injuries. Neurosurgery. 2001;48(6):1255–1258.
27. Stein SC, Spettell C. The Head Injury Severity Scale (HISS): a prac-tical classification of closed-head injury. Brain Inj. 1995;9(5):437–444.
28. Stein SC. Minor head injury: 13 is an unlucky number. J Trauma.2001;50(4):759–760.
29. Miller EC, Derlet RW, Kinser D. Minor head trauma: is computedtomography always necessary? Ann Emerg Med. 1996;27(3):290–294.
30. Muller K, Elverland A, Romner B, et al. Analysis of protein S-100B in serum: a methodological study. Clin Chem Lab Med.2006;44(9):1111–1114.
31. Lee H, Wintermark M, Gean AD, Ghajar J, Manley GT, Mukher-jee P. Focal lesions in acute mild traumatic brain injury andneurocognitive outcome: CT versus 3T MRI. J Neurotrauma.2008;25(9):1049–1056.
32. Doezema D, King JN, Tandberg D, Espinosa MC, Orrison WW.Magnetic resonance imaging in minor head injury. Ann EmergMed. 1991;20(12):1281–1285.
33. Topal NB, Hakyemez B, Erdogan C, et al. MR imaging in thedetection of diffuse axonal injury with mild traumatic brain injury.Neurol Res. 2008;30(9):974–978.
34. Muller K, Townend W, Biasca N, et al. S100B serum level predictscomputed tomography findings after minor head injury. J Trauma.2007;62(6):1452–1456.
35. Heneghan C. EBM resources on the new CEBM website. EvidBased Med. 2009;14(3):67.
36. Whiting PF, Weswood ME, Rutjes AW, Reitsma JB, BossuytPN, Kleijnen J. Evaluation of QUADAS, a tool for the qualityassessment of diagnostic accuracy studies. BMC Med Res Methodol.2006;6:9.
37. Ingebrigtsen T, Waterloo K, Jacobsen EA, Langbakk B, Romner B.Traumatic brain damage in minor head injury: relation of serumS-100 protein measurements to magnetic resonance imaging andneurobehavioral outcome. Neurosurgery. 1999;45(3):468–475.
38. Ingebrigtsen T, Romner B, Marup-Jensen S, et al. The clinicalvalue of serum S-100 protein measurements in minor head injury:a Scandinavian multicentre study. Brain Inj. 2000;14(12):1047–1055.
39. Poli-de-Figueiredo LF, Biberthaler P, Simao Filho C, Hauser C,Mutschler W, Jochum M. Measurement of S-100B for risk classi-fication of victims sustaining minor head injury—first pilot studyin Brazil. Clinics (Sao Paulo). 2006;61(1):41–46.
40. Mussack T, Biberthaler P, Kanz KG, et al. Immediate S-100B andneuron-specific enolase plasma measurements for rapid evalua-tion of primary brain damage in alcohol-intoxicated, minor head-injured patients. Shock. 2002;18(5):395–400.
41. Biberthaler P, Mussack T, Wiedemann E, et al. Rapid identificationof high-risk patients after minor head trauma (MHT) by assess-ment of S-100B: ascertainment of a cut-off level. Eur J Med Res.2002;7(4):164–170.
42. Biberthaler P, Mussack T, Wiedemann E, et al. Evaluation of S-100b as a specific marker for neuronal damage due to minor headtrauma. World J Surg. 2001;25(1):93–97.
43. Nygren De Boussard C, Fredman P, Lundin A, Andersson K, Ed-man G, Borg J. S100 in mild traumatic brain injury. Brain Inj.2004;18(7):671–683.
44. Bazarian JJ, Beck C, Blyth B, von Ahsen N, Hasselblatt M. Im-pact of creatine kinase correction on the predictive value of S-100B after mild traumatic brain injury. Restor Neurol Neurosci.2006;24(3):163–172.
45. Morochovic R, Racz O, Kitka M, et al. Serum S100B protein inearly management of patients after mild traumatic brain injury.Eur J Neurol. 2009;16(10):1112–1117.
46. Bouvier D, Oddoze C, Ben Haim D, et al. Interest of S100Bprotein blood level determination for the management of patientswith minor head trauma. Ann Biol Clin (Paris). 2009;67(4):425–431.
47. Oh EJ, Kim YM, Jegal DW, Kahng J, Park YJ, Han K. Diagnosticvalue of Elecsys S100 as a marker of acute brain injury in theemergency department. J Clin Lab Anal. 2007;21(6):387–392.
48. Jagoda AS, Bazarian JJ, Bruns JJ Jr, et al. American College ofEmergency Physicians; Centers for Disease Control and Preven-tion. Clinical policy: neuroimaging and decision making in adultmild traumatic brain injury in the acute setting. Ann Emerg Med.2008;52(6):714–748.
49. Brismar B, Engstrom A, Rydberg U. Head injury and intoxi-cation: a diagnostic and therapeutic dilemma. Acta Chir Scand.1983;149(1):11–14.
50. Sloan EP, Zalenski RJ, Smith RF, et al. Toxicology screening inurban trauma patients: drug prevalence and its relationship totrauma severity and management. J Trauma. 1989;29(12):1647–1653.
51. Enochsson L, Carlsson-Sanz S, van der Linden J. The influenceof alcohol and time on the S-100B levels of patients with minorhead injury. Eur J Neurol. 2005;12(6):445–448.
52. Biberthaler P, Mussack T, Wiedemann E, et al. Elevated serumlevels of S-100B reflect the extent of brain injury in alcohol intox-icated patients after mild head trauma. Shock. 2001;16(2):97–101.
53. Kuppermann N, Holmes JF, Dayan PS, et al. Pediatric Emer-gency Care Applied Research Network (PECARN). Identifica-tion of children at very low risk of clinically-important braininjuries after head trauma: a prospective cohort study. Lancet.2009;374(9696):1160–1170.
54. Castellani C, Stojakovic T, Cichocki M, et al. Reference ranges forneuroprotein S-100B: from infants to adolescents. Clin Chem LabMed. 2008;46(9):1296–1299.
55. Spinella PC, Dominguez T, Drott HR, et al. S-100beta protein-serum levels in healthy children and its association with outcomein pediatric traumatic brain injury. Crit Care Med. 2003;31(3):939–945.
56. Gazzolo D, Michetti F, Bruschettini M, et al. Pediatric concen-trations of S100B protein in blood: age- and sex-related changes.Clin Chem. 2003;49(6, Pt 1):967–970.
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
www.headtraumarehab.com
240 JOURNAL OF HEAD TRAUMA REHABILITATION/JULY–AUGUST 2010
57. Berger RP, Beers SR, Richichi R, Wiesman D, Adelson PD. Serumbiomarker concentrations and outcome after pediatric traumaticbrain injury. J Neurotrauma. 2007;24(12):1793–1801.
58. Berger RP, Dulani T, Adelson PD, Leventhal JM, Richichi R,Kochanek PM. Identification of inflicted traumatic brain injuryin well-appearing infants using serum and cerebrospinal markers:a possible screening tool. Pediatrics. 2006;117(2):325–332.
59. Berger RP, Pierce MC, Wisniewski SR, Adelson PD, KochanekPM. Serum S100B concentrations are increased after closedhead injury in children: a preliminary study. J Neurotrauma.2002;19(11):1405–1409.
60. Berger RP, Adelson PD, Pierce MC, Dulani T, Cassidy LD,Kochanek PM. Serum neuron-specific enolase, S100B, and myelinbasic protein concentrations after inflicted and noninflicted trau-matic brain injury in children. J Neurosurg. 2005;103(1) (suppl):61–68.
61. Bechtel K, Frasure S, Marshall C, Dziura J, Simpson C. Relation-ship of serum S100B levels and intracranial injury in children withclosed head trauma. Pediatrics. 2009;124(4):e697–e704.
62. Castellani C, Bimbashi P, Ruttenstock E, Sacherer P, Stojakovic T,Weinberg AM. Neuroprotein s-100B—a useful parameter in pae-diatric patients with mild traumatic brain injury? Acta Paediatr.2009;98(10):1607–1612.
63. Berger RP, Kochanek PM. Urinary S100B concentrations are in-creased after brain injury in children: a preliminary study. PediatrCrit Care Med. 2006;7(6):557–561.
64. Gazzolo D, Marinoni E, Di Iorio R, et al. Urinary S100B pro-tein measurements: a tool for the early identification of hypoxic-ischemic encephalopathy in asphyxiated full-term infants. CritCare Med. 2004;32(1):131–136.
65. Pickering A, Carter J, Hanning I, Townend W. Emergency depart-ment measurement of urinary S100B in children following headinjury: can extracranial injury confound findings? Emerg Med J.2008;25(2):88–89.
66. Schultke E, Sadanand V, Kelly ME, Griebel RW, Juurlink BH. Canadmission S-100beta predict the extent of brain damage in headtrauma patients? Can J Neurol Sci. 2009;36(5):612–616.
67. Wiesmann M, Steinmeier E, Magerkurth O, Linn J, Gottmann D,Missler U. Outcome prediction in traumatic brain injury: com-parison of neurological status, CT findings, and blood levels ofS100B and GFAP. Acta Neurol Scand. 2009;5:178–185.
68. Lumpkins KM, Bochicchio GV, Keledjian K, Simard JM, McCunnM, Scalea T. Glial fibrillary acidic protein is highly correlated withbrain injury. J Trauma. 2008;65(4):778–782.
69. Pelinka LE, Kroepfl A, Leixnering M, Buchinger W, Raabe A,Redl H. GFAP versus S100B in serum after traumatic brain in-
jury: relationship to brain damage and outcome. J Neurotrauma.2004;21(11):1553–1561.
70. Pelsers MM, Hanhoff T, Van der Voort D, et al. Brain- and heart-type fatty acid-binding proteins in the brain: tissue distributionand clinical utility. Clin Chem. 2004;50(9):1568–1575.
71. Yamazaki Y, Yada K, Morii S, Kitahara T, Ohwada T. Diagnos-tic significance of serum neuron-specific enolase and myelinbasic protein assay in patients with acute head injury. Surg Neurol.1995;43(3):267–270.
72. de Kruijk JR, Leffers P, Menheere PP, Meerhoff S, Twijnstra A.S-100B and neuron-specific enolase in serum of mild traumaticbrain injury patients. A comparison with health controls. ActaNeurol Scand. 2001;103(3):175–179.
73. Pineda JA, Lewis SB, Valadka AB, et al. Clinical significance ofalphaII-spectrin breakdown products in cerebrospinal fluid af-ter severe traumatic brain injury. J Neurotrauma. 2007;24(2):354–366.
74. Pelinka LE, Kroepfl A, Schmidhammer R, et al. Glial fibrillaryacidic protein in serum after traumatic brain injury and multipletrauma. J Trauma. 2004;57(5):1006–1012.
75. Haydel MJ. Clinical decision instruments for CT scanning in mi-nor head injury. JAMA. 2005;294(12):1551–1553.
76. Corwin MT, Donohoo JH, Partridge R, Egglin TK, Mayo-SmithWW. Do emergency physicians use serum D-dimer effectivelyto determine the need for CT when evaluating patients for pul-monary embolism? Review of 5,344 consecutive patients. AJRAm J Roentgenol. 2009;192(5):1319–1323.
77. Wells PS, Owen C, Doucette S, Fergusson D, Tran H. Does thispatient have deep vein thrombosis? JAMA. 2006;295(2):199–207.
78. Keller T, Zeller T, Peetz D, et al. Sensitive troponin I assay inearly diagnosis of acute myocardial infarction. N Engl J Med.2009;361(9):868–877.
79. Unden J, Bellner J, Astrand R, Romner B. Serum S100B levels inpatients with epidural haematomas. Br J Neurosurg. 2005;19(1):43–45.
80. Townend W, Dibble C, Abid K, Vail A, Sherwood R, Lecky F.Rapid elimination of protein S-100B from serum after minor headtrauma. J Neurotrauma. 2006;23(2):149–155.
81. Romner B, Ingebrigtsen T, Kongstad P, Børgesen SE. Traumaticbrain damage: serum S-100 protein measurements related to neu-roradiological findings. J Neurotrauma. 2000;17(8):641–647.
82. Ruan S, Noyes K, Bazarian JJ. The economic impact of S-100B asa pre-head CT screening test on emergency department manage-ment of adult patients with mild traumatic brain injury. J Neuro-trauma. 2009;26(10):1655–1664.
Copyright © 2010 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.