1

Full title page

Intraventricular Hemorrhage on Initial Computed Tomography as Marker

of Diffuse Axonal Injury after Traumatic Brain Injury

Regular manuscript

Daddy Mata-Mbemba1, MD, PhD; Shunji Mugikura

1, MD, PhD; Atsuhiro

Nakagawa2, MD, PhD; Takaki Murata

1, MD, PhD; Yumiko Kato

1, MD,

PhD; Yasuko Tatewaki1, MD, PhD; Li Li

1 MD, PhD; Kei Takase

1, MD,

PhD; Kiyoshi Ishii, MD, PhD Shigeki Kushimoto3, MD, PhD; Teiji

Tominaga2, MD, PhD, and Shoki Takahashi

1, MD, PhD.

Department of 1Diagnostic Radiology,

2 Neurosurgery, and

3Division of

Emergency Medicine, Tohoku University Graduate School of Medicine,

Sendai, Japan and

4Department of Radiology, Sendai city hospital, Sendai, Japan

*Address correspondence to:

Shunji Mugikura, M.D., Ph.D.

Department of Diagnostic Radiology, Tohoku University, Graduate School

of Medicine, 1-1 Seiryo-machi, Aoba-ku, Sendai, Japan 980-8574

Tel: International access code +81-22-717-7312

Fax: International access code +81-22-717-7316

Email: mugi@rad.med.tohoku.ac.jp

Page 1 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

2

Abbreviated title page

Intraventricular Hemorrhage on Initial Computed Tomography as Marker

of Diffuse Axonal Injury after Traumatic Brain Injury

Page 2 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

3

ABSTRACT:

Intraventricular hemorrhage (IVH) on initial Computed tomography (CT) was reported

to predict lesions of diffuse axonal injury (DAI) in the corpus callosum on subsequent

magnetic resonance Imaging (MRI). We aimed to examine the relationship between

initial CT findings and DAI lesions detected on MRI, and the relationship between the

severity of IVH (IVH score) and severity of DAI (DAI staging). A consecutive 140

patients with traumatic brain injury (TBI) who underwent MRI within 30 days after

onset were revisited. We reviewed their initial CT for the following six findings: status

of basal cistern, status of midline shift, epidural hematoma, IVH, subarachnoid

hemorrhage, and volume of hemorrhagic mass and IVH score was assigned in each

patient. Based on MRI findings, patients were divided into DAI and non-DAI groups,

and were assigned a DAI staging. Then, to confirm the IVH on initial CT predicts DAI

lesions on MRI, we used multivariate analysis of the six CT findings including IVH and

examined the relationship between IVH score and DAI staging. The IVH detected on

CT was the only predictor of DAI (P=0.0139). The IVH score and DAI staging showed

significant positive correlation (P

4

conclusion, IVH on initial CT is the only maker of DAI on subsequent MRI, specifically

severe DAI (stage 2 or 3)

Key words

computed tomography

intraventricular hemorrhage

diffuse axonal injury

traumatic brain injury

corpus callosum

Page 4 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

5

INTRODUCTION:

Diffuse axonal injury (DAI) is thought to be one of the most important factors in

determining the prognosis of patients with traumatic brain injury (TBI).1 For instance, a

severe DAI may transform young, productive individuals into dependent patients,

sometimes requiring institutionalized care, which results in increased socio-economic

burden .2-3

In the routine practice, CT remains the first choice imaging modality for initial

and early follow-up evaluations in patients with TBI,4 while, MRI is often additionally

performed later mainly to screen for lesions suggesting DAI (DAI lesions), because CT

fails to accurately reveal them. In that sense, knowing the initial CT findings predicting

DAI lesions on subsequent MRI will assist emergency physicians to properly select

patients who should undergo MRI study.

The three-grading of DAI based on the locations of DAI lesions on MRI (DAI

stage) as proposed by Gentry et al.5

was reported to significantly predict the functional

outcome of patients with the stage 3 being the worst.6-7

Recently, Matsukawa et al.8

reported that the presence or severity of intraventricular hemorrhage (IVH) on initial CT

was associated with DAI lesions in the corpus callosum on subsequent MRI (DAI stage

Page 5 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

6

2), however they did not examine the relationship between IVH and DAI lesions located

in the cerebral hemisphere (DAI stage 1) or those in the brain stem (DAI stage 3)..

Therefore, to clarify the clinical significance of IVH on initial CT for predicting

the DAI lesions on subsequent MRI, we sought to examine: 1) the relationship between

various initial CT findings, including IVH, and the presence of DAI on subsequent MR

imaging, and 2) the relationship between the severity of IVH on initial CT and stage of

DAI on MRI, which has never been investigated to the best of our knowledge.

PATIENTS AND METHODS:

Study population

We retrospectively reviewed the records of all consecutive TBI patients who

were registered in the electronic radiological database of our institution, a major tertiary

referral hospital in northeastern japan, between the January 2007 and June 2011. One

hundred-forty patients (age range: 6-89 years) who underwent initial MRI within 30

days after onset were included in this study.6-7, 9

Male sex was predominant (102

patients [72.9%]) and Traffic accident was the main mean of injury (94 patients

[67.1%]). At the time of their admission to the emergency room, clinical symptoms

were mild (Glasgow coma scale [GCS] GCS 1315) in 89(63.9%) patients, moderate

Page 6 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

7

(GCS 9-12) in 27 (19.3%) patients, and severe in 16 (11.4%) patients. Eight patients

were missing a GCS data. In all of the 140 patients, the initial CT was performed within

the 24 hours following the onset.

Approval was obtained from the institutional research ethic board, and

informed consent from patients was waived.

Clinical and demographic Data

In our institution, the picture archiving and communications system (PACS)

contains the structured order entry system in which the referring emergency physicians

are requested to put relevant clinical data of the patients. Therefore, the radiologists are

provided with Glasgow coma scale (GCS) at admission and the age of the patients,

which have been reported to be the two powerful clinical and demographic data in

predicting outcome of TBI patient .10

The sex of the patient, the mean of injury and the

time at admission are also provided. The interval between admission and initial MRI

was also calculated. Therefore, in this study, the following 5 demographic data are

briefly summarized in the background analysis of our patients: GCS at admission, age,

sex, mean of injury, and the time interval from admission to initial MRI.

Page 7 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

8

CT evaluation

In TBI patients, Rotterdam and Marshall CT scoring systems have been widely

used for the purpose of classification and outcome prediction .9, 11, 12, 13

Indeed, recent

studies have documented that the both CT scoring systems can accurately predict death

at hospital discharge or a long-term functional outcome in TBI patients.11, 14, 15, 16

In that

sense, in the current study, we evaluated initial CT obtained within 24 hours after the

onset regarding the following findings sufficiently validated either in the Marshall or

Rotterdam CT scoring system :11, 14, 15, 16

basal cistern status, presence of midline shift,

presence of epidural hematoma (EDH), volume of the hemorrhagic mass, and presence

of either intraventricular hemorrhage or subarachnoid hemorrhage or both (IVH/SAH).

In order to clarify the relationship between IVH or SAH and DAI, the two items (IVH

and SAH) were assessed in two different settings: 1) combined in one variable of

IVH/SAH as included in Rotterdam score, 11, 14 and 2) separated into two different

variables of IVH and SAH. 7, 8, 9, 17

Furthermore, we graded the severity of the IVH (IVH score, range 08)

following Matsukawa et al:8 the patients score represented a sum of the number of

ventricles in which IVH was seen; anterior horns [unilateral or bilateral], posterior horns

Page 8 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

9

[unilateral or bilateral], inferior horns [unilateral or bilateral], third ventricle, and fourth

ventricle. Patient who did not show IVH on CT were scored 0.

Blinded to a clinical situation and MRI findings of patients, two neuroradiologists

(D.M.M & S. M) independently reviewed initial CT for following six items: basal

cistern status, presence of midline shift, EDH, volume of the hemorrhagic mass, IVH,

and SAH. Consensus was used to solve disagreements between readers.

Basal cistern status was classified as normal, compressed, or absent. Midline shift

was defined as displacement of the septum pellucidum in relation to the midline, and

was recorded in millimeters. A midline shift > 5 mm was scored as present and a shift

5 mm was scored as absent.11, 12, 14

The overall volume of each hemorrhagic mass was

calculated by digital measurement using a dedicated workstation by multiplying the sum

area of hemorrhagic masses on each slice by the slice thickness.14, 18

The volume of

hemorrhagic mass in each patient was graded as absent,

10

T2-weighted imaging, axial spin echo T1-weighted imaging, axial fluid-attenuated

inversion recovery (FLAIR) imaging, axial T2*-weighted gradient echo imaging

(T2*WI), Diffusion weighted imaging (DWI), and MR angiography using time-of-flight

sequence.

Evaluation of MRI

Blinded to clinical situation and CT findings of patients; another group of two

neuroradiologists (T.M & Y.T) reviewed the initial MRI for the presence of DAI using

DWI, T2WI, FLAIR, and T2*WI. Consensus was used to solve disagreements between

readers.

A hemorrhagic DAI was defined as hypointense foci noted on T2*WI, that was

not compatible with vascular, bony, or artifactual structures and was located in the

consistent brain regions.19

Lesions in the cerebral cortex were not defined as DAI, rather

as contusions.7, 9

A nonhemorrhagic DAI was defined as hyperintense focus noted on

DWI, FLAIR or T2WI, which were located in the consistent brain regions without

association of hypointense foci on the corresponding T2*WI.19, 20

Based on the MRI results, we classified all 140 patients into following two

settings: (a) DAI versus non-DAI groups. The DAI group encompassed hemorrhagic

and nonhemorrhagic DAI.7, 13

(b) Patients were assigned one of the 4 DAI staging

Page 10 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

11

proposed by Gentry et al:5 stage 0: representing non-DAI, 1: DAI lesions confined to

the lobar white matter or cerebellum, 2: DAI lesions located in the corpus callosum with

or without lesions of stages 1, and 3 DAI lesions located in the brain stem with or

without lesions of stages 1 and/or 2.

Analysis and statistics

First of all, we performed a background analysis to determine the relationship

between patients demographic data and the presence of DAI using univariate and

multivariate logistic regressions.

Next, to test the hypothesis that the presence of IVH on initial CT can predict the

overall DAI disease on subsequent MRI, we examined the relationship between all six

initial CT findings, including IVH, and the presence of DAI on MRI by using univariate

and multivariate logistic regressions. These statistic tests were conducted twice, that is

we evaluated the performance of IVH using two different settings: 1) one variable of

IVH/SAH including either IVH or SAH or both, and 2) separated into two variables

of IVH and SAH.

Finally, when the IVH on initial CT was proved to predict the presence of DAI

on subsequent MRI, we further tested whether there is a positive correlation between the

IVH score and DAI stage using Spearman rank correlation and examined whether there

Page 11 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

12

is any significant difference in the IVH score between DAI score 0 (considered as

baseline) and DAI stage 1, 2, and 3 by using a nonparametric comparisons with control

using Steel method.

Furthermore, in order to determine if the statistical associations observed in our

study have clinical implications or not, we calculated the sensitivity, specificity, positive

and negative predictive values of the IVH detected on CT in predicting DAI on

subsequent MRI.

All statistical analyses were performed by using software JMP Pro version 10

(SAS Institute, Inc., Cary, North Carolina) and P value below 0.05 was considered

statistically significant.

RESULTS:

DAI detected by MRI

Of these 140 patients, 48 patients (34, 3%) were assigned to DAI group and 92

patients (65.7%) to non-DAI group based on MR imaging. According to the stage of

DAI; 92 patients (65.7%) had stage 0, 22 patients (15.7%) had stage 1, 13 patients (9.3)

had stage 2, and 13 patients (9.3%) has stage 3. All of our patients (100%) with DAI

stage 2 (corpus callosum lesions) had also lesion in the lobar white matter or cerebellum.

Of 13patient with DAI stage 3, 10 (76.9%) had additional lesions in the lobar white

Page 12 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

13

matter or cerebellum and corpus callosum lesions, while the remaining patients showed

DAI lesions located in the brain stem in one patient, brainstem and corpus callosum in

one, and brainstem and lobar white matter in one.

The majority [29 (61.7 %) of 48 patients)] of DAI patients showed coexistence

of hemorrhagic and nonhemorrhagic DAI lesions. Hemorrhagic DAI alone were noted

in17 patients (34%), whereas nonhemorrhagic DAI alone were detected only in 2

patients (4.3%).

Patients demographics and DA I

The demographic data in DAI or non-DAI groups are shown in the table 1. Male

sex and GCS were significantly associated with DAI both in univariate (Male,

P=0.0169; GCS, P=0.0004) and multivariate (Male, OR=2.9, P=0.0329; GCS OR=9.9,

P=0.004) analyses. No significant difference was noted between DAI and non-DAI

groups about the age, the means of injury, and the interval of time between

accident-and-initial MR scan of patients.

Relationship between initial CT findings and presence of DAI on subsequent MRI

In our study, the IVH was seen in 27 patients (19%), among whom only 1

patient showed an isolated IVH (without any associated TBI lesions). The remaining 26

(96.3% of 27) patients had IVH associated at least with SAH on CT.

Page 13 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

14

In the setting of IVH combined with SAH (IVH/SAH) in a single variable

(Annexed table); the items of IVH/SAH was significantly associated with DAI on

univariate logistic regression (P=0.0290). However, on multivariate logistic regression

when adjusted with other initial CT findings (status of basal cistern, status of midline

shift, epidural hematoma, and volume of hemorrhagic mass), none of these items

including IVH/SAH (P=0.1120) was significantly associated with DAI.

Whereas in the setting of IVH and SAH separated in two independent

variables (Table 2), IVH detected on CT was the only predictor of DAI on both

univariate (OR: 3.7, P= 0.0034) and multivariate (P=0.0139, adjusted OR: 4.2; 95%CI:

1.3-14.3) logistic regressions.

Relationship between The IVH score and the staging of DAI

The IVH score and the staging of DAI showed significant positive correlation

(0.3005, P

15

The presence of IVH on initial CT had a sensitivity of 33, 3%; specificity of

88%, positive predictive value of 59.3% and negative predictive value of 71.7% in

predicting DAI on MRI.

DISCUSSION

In this retrospective study, we have sought to examine the relationship between

the initial CT finding and DAI lesions detected on subsequent MRI in patients with TBI

and we found that only the presence of IVH on initial CT predicts the DAI lesions.

Furthermore, we demonstrated that the higher the IVH score, the higher

becomes the stage of DAI, and IVH score in severe DAI (stage 2 or 3) was significantly

higher than that of the DAI stage 0. This indicates that using the IVH score may predict

the risk of having severe DAI.

From their experimental works, Holbourn et al. 21, 22

reported that the

post-traumatic brain lesions occur by two major mechanisms: direct injuries (contact

phenomena to the skull) and indirect injuries (shearing strain) that may arise irrespective

of skull deformation. The shearing strain that consists of angular or rotational

acceleration - deceleration forces, leads to the stretching and the deformation of the

brain tissue including vessels and axons.23, 24

This mechanism is thought to be the most

Page 15 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

16

responsible for production of DAI.8, 21, 25

Besides, the shearing strain has been

incriminated to act maximally at parasagittal structures;26

mostly at the junction of the

corpus callosum with the septum pellucidum and fornix, as well as on the ventricle wall,

thereby damaging subependymal vessels, hence produces IVH.8, 25, 26

Thus, the present

observation that IVH detected on initial CT predicts DAI on subsequent MRI can be

explained by our assumption that both lesions (IVH and DAI) occur mainly from the

same etiological mechanism of shearing strain.

Moreover, we demonstrated that the higher the IVH score, the higher becomes

the stage of DAI, specifically severe DAI (stage 2 or 3). Indeed, corpus callosum and

the brainstem are also structures located in the midline of the brain. As shearing strain

acts maximally in the parasagittal regions, we believe that these two structures are

concomitantly injured with subependymal vessels that produce IVH. That is, with

greater shearing strain, proportionally and simultaneously more IVH and DAI lesions in

midline brain structures (corpus callosum and midbrain) are produced (Figure 2).8

This

implies that the presence of IVH on initial CT should be considered as marker of

severer DAI (stage 2 and 3) and warrant DAI workup.

In contrast to the shearing strain mechanism, the direct injuries that consists of

the head collision, is mainly incriminated to produce lesions that result in increased

Page 16 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

17

intracranial pressure including any voluminous hemorrhagic mass, compression of basal

cisterns and midline shift.9, 14

In our study, none of these lesions was significantly

associated with DAI. In accord with this result, Adam et al. reported significantly lower

increased intracranial pressure in DAI patients.27

Similarly, Lee et al. demonstrated that

monitoring of the intracranial pressure is not needed in DAI patients when no

hemorrhagic mass was associated.28

Therefore, the lack of significant association

between DAI and status of basal cistern, positive midline shift, EDH or volume of

hemorrhagic mass shown in our data, is probable due to the difference in their

etiological mechanisms, in that, DAI come mainly from the shearing strain whereas

status of basal cistern, positive midline shift, EDH and volume of hemorrhagic mass are

mainly produced by direct injuries. Actually, there is no need to say that in very

severe TBI; both mechanisms can be involved, resulting to the mixture of their

respective lesions.

Our results reporting lack of association between voluminous hemorrhagic mass

and DAI opposes to the findings of Skandsen et al.7 who reported a significant

association between DAI and voluminous hemorrhagic mass. Their study included TBI

patients from a trauma-specialized hospital where severe cases are transferred and mild

cases were excluded in their series. Therefore, the association between voluminous

Page 17 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

18

hemorrhagic mass and DAI reported by these authors could have been affected by the

fact that their patients were severely injured. This can be understood by the higher

prevalence (72%) of DAI in their series. Indeed, according to the other pathologically

studies, DAI were detected in almost all fatal cases, and often in association with

voluminous hemorrhagic mass.2 Whereas, in our study including considerable number

of mild cases (89 patients [63.9%]) with less severe cases (16 patients [11.4%]), the

prevalence of DAI was less than half (34, 3%) of the report by Skandsen et al. In our

opinion, the spectrum of the severity of TBI patients in our study is consistent with the

literature which reports about 10% the prevalence of severe TBI in developed countries.

In that sense, our results could be closer to the daily clinical context.

Lagares et al. also found association between IVH and DAI. However, they

used only univariate analyses.29

Because TBI lesions often coexist, we believe that

multiple logistic regressions constitute the statistic method of choice in order to

elucidate confounding factors among potential independent predictors.

Recently Matsukawa et al documented that IVH on initial CT significantly

associated with DAI lesions located in the corpus callosum, however, their study did not

include patients with DAI lesions in lobar white matter or cerebellum, or brain stem.8

Page 18 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

19

In the light of our data in which the majority (76.9%) of patients with brainstem

DAI lesions (stage 3) had also DAI lesions in lobar white matter or cerebellum, or brain

stem; in this study, where we included all consecutives DAI patients regardless of the

location of DAI lesions, we have further shown that the IVH on initial CT predicts DAI

irrespective of their location, and we have also demonstrated that using the IVH score

may predict the risk of having severe DAI (stage 2 or 3), that is not only stage 2 but also

stage 3, which has never been investigated to the best of our knowledge. We believe that

this information could be useful for clinician, in that, it will allow a proper selection of

patients who should further undergo MRI. That is, when a given patient shows IVH on

initial CT, MRI study should be performed as workup for DAI disease.

In our study, the IVH on CT yielded a high specificity (88%) and negative

predictive value (71.7%) but a relatively low sensitivity (33.3%) in predicting DAI on

subsequent MRI. The prevalence of IVH on initial CT in our study (19. 3 %) was

similar than that reported in the recent literature.9, 10, 11

Of the 27 patients who showed

IVH on CT in our study, 26 (96.3%) had coexistence of IVH and SAH. This finding

allowed us to speculate that some of IVH seen in our study, specifically in patients

scanned later after the onset, could be secondary IVH representing a reflux of SAH via

the outlet foramina of the fourth ventricle rather than primary IVH. In that sense, the

Page 19 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

20

relatively low sensitivity (33, 3%) of IVH in predicting DAI on subsequent MRI noted

in this study could be explained by a possible blurring effect of a secondary IVH on the

true sensitivity of the primary IVH. However, the high specificity (88%) and negative

predicting value (71.7%) documented in our study suggest that in clinical routine, IVH

should be considered as a surrogate of shearing strain. That is, as we assume that

shearing strain produces proportionally and concomitantly IVH and DAI, absence of

IVH on CT suggests that shearing strain is not the main mechanism of TBI, thus, DAI

will less likely be seen on subsequent MRI.

In our series, only one patient showed an isolated IVH. Such extremely rare

prevalence of isolated IVH is in accord with a large scale studies that included 8,374

and 5000 patients with TBI, in which isolated IVH were seen in only 8 (0.09%) and 6

(0.1%) patients, respectively.30-31

In contrast to the rarity of isolated IVH; in our series,

SAH was sometimes seen on CT in patients with DAI and on univariate analysis, it

showed a trend (P=0.0512) of association with DAI. Based on our findings and on the

hypothesis stipulating that significant force is required to produce traumatic IVH, 32, 33

we think that some type of SAH and IVH are caused by the same mechanism of

shearing strain: the former appears with a lower energy whereas the latter with higher

Page 20 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

21

energies. A dedicated study should be conducted to reveal a shearing-strain related SAH

that could warrant a DAI work-up, as does IVH.

We should acknowledge some limitations in our study. In this retrospective

designs, not all TBI patients admitted to the emergency room were studied using MRI.

This study included only the consecutive patients who underwent MRI without definite

standardized clinical indication of MRI. Moreover, detectability of nonhemorrhagic

DAI on DWI or conventional MR images tend to decrease over the time, thus, some

nonhemorrhagic lesions could have not been detected in patients examined later after

the onset of TBI.

CONCLUSION: After TBI, presence of IVH on initial CT should be considered as

maker of severe DAI and warrant appropriated DAI work-up.

Page 21 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

22

References:

1. Li XY, and Feng DF. (2009). Diffuse axonal injury: novel insight into detection and

treatment. J Clin Neurosci 16, 614-619.

2. Gentry LR. (1994). Imaging of closed head injury (review). Radiology 191:1-17.

3. Strich SJ. (1956). Diffuse degeneration of cerebral white matter in severe dementia

following head injury. J Neurol Neurosurg Psychiatry 19:163-185

4. James M, and Provenzale. (2010). Imaging of traumatic brain injury: a review of the

recent medical literature. AJR 194:1619.

5. Gentry LR, Godersky JC, and Thompson B. (1988) MR imaging of head trauma:

review of the distribution and radiopathologic features of traumatic lesions. AJR 150,

663-672.

6. Hilario A, Ramos A, Millan JM, Salvador E, Gomez PA, Cicuendez M, Diez-Lobato

R, and Lagares A. (2012). Severe traumatic head injury: prognostic value of brain

stem injuries detected at MRI. AJNR Am J Neuroradiol 33, 1925-1931.

7. Skandsen T1, Kvistad KA, Solheim O, Strand IH, Folvik M, and Vik A. (2010).

Prevalence and impact of diffuse axonal injury in patients with moderate and severe

head injury: a cohort study of early magnetic resonance findings and 1-year outcome.

J Neurosurg 113, 556-563.

Page 22 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

23

8. Matsukawa H, Shinoda M, Fujii M, Takahashi O, Murakata A, Yamamoto

D, Sumiyoshi S, and Ishikawa R. (2012). Intraventricular hemorrhage on computed

tomography and corpus callosum injury on magnetic resonance imaging in patients

with isolated blunt traumatic brain injury. J Neurosurg 117; 334-339

9. Lagares A1, Ramos A, Prez-Nuez A, Ballenilla F, Alday R, Gmez PA, Kaen

A, and Lobato RD.(2009). The role of MR imaging in assessing prognosis after

severe and moderate head injury. Acta Neurochir (Wien) 151, 341356

10. Nelson DW1, Nystrm H, MacCallum RM, Thornquist B, Lilja A, Bellander

BM, Rudehill A, Wanecek M, and Weitzberg E. (2010). Extended analysis of early

computed tomography scans of traumatic brain injured patients and relations to

outcome. J. Neurotrauma 27, 51-64.

11. Maas AI1, Hukkelhoven CW, Marshall LF, and Steyerberg EW. (2005). Prediction

of outcome in traumatic brain injury with computed tomographic characteristics: a

comparison between the computed tomographic classification and combination of

computed tomographic predictors. Neurosurgery 57, 1173-1182.

12. Marshall LF1, Marshall SB, Klauber MR, Van Berkum Clark M, Eisenberg H, Jane

JA, Luerssen TG, Marmarou A, and Foulkes MA. (1991). A new classification of

head injury based on computerized tomography. J Neurosurgery; 75(Suppl),

Page 23 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

24

S14-S20.

13. Chelly H, Chaari A, Daoud E, Dammak H, Medhioub F, Mnif J, Hamida

CB, Bahloul M, and Bouaziz M. (2011). Diffuse axonal injury in patients with head

injuries: an epidemiologic and prognostic study of 124 cases. J Trauma 71(4),

838-846.

14. Mata-Mbemba D, Mugikura S, Nakagawa A, Murata T, Ishii K, Li L, Takase K,

Kushimoto S, and Takahashi S. (2014). Early CT findings to predict early death at

hospital discharge in patients with traumatic brain injury: Marshall and Rotterdam

CT scoring systems compared in the tertiary referral hospital in northeastern Japan.

Acad Radiol 21:605-611

15. Katsnelson M1, Mackenzie L, Frangos S, Oddo M, Levine JM, Pukenas B, Faerber

J, Dong C, Kofke WA, and le Roux PD. (2012). Are initial radiology and clinical

scales associated with subsequent intracranial pressure and brain oxygen level after

severe traumatic brain injury. Neurosurgery 70(5),1095-1105.

16. Huang YH, Deng YH, Lee TC, and Chen WF. (2012). Rotterdam computed

tomography score as a prognosticator in head-injured patients undergoing

decompressive craniectomy. Neurosurgery 71(1), 80-85.

17. Adam JH., Graham DI., Murray LS, and Scout G. (1982). Diffuse axonal injury due

Page 24 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

25

to nonmissile head injury in humans: an analysis of 45 cases. Ann Neurol 12,

557-563.

18. Jacobs B1, Beems T, van der Vliet TM, Diaz-Arrastia RR, Borm GF, and Vos PE.

(2011). Computed Tomography and outcome in moderate and severe traumatic brain

injury: hematoma volume and midline shift revisited. J Neurotrauma 28(2),

203-215.

19. Ezaki Y, Tsutsumi K, Morikawa M, and Nagata I. (2006). Role of

diffusion-weighted magnetic resonance imaging in diffuse axonal injury. Acta

Radiol 47, 733-740.

20. Tong KA, Ashwal S, Holshouser BA, Nickerson JP, Wall CJ, Shutter LA, Osterdock

RJ, Haacke EM, and Kido D. (2004). Diffuse axonal injury in children: clinical

correlation with hemorrhagic lesions. Ann Neurol 56, 36-50.

21. Holbourn AHS (1943). Mechanics of head injuries. Lancet 2, 438-441.

22. Holbourn AHS. (1945). The mechanics of brain injuries. Br Med Bull 3, 147-149.

23. Ommaya AK, Goldsmith W, and Thibault L. (2002). Biomechanics and

neuropathology of adult and pediatric head injury. Br J Neurosurg 16, 220-242.

24. Werner C. and Engelhard K. (2007). Pathophysiology of traumatic brain injury. Br J

Anaesth 99, 4-9.

Page 25 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

26

25. Orrison WW, Gentry LR, Stimac GK, Tarrel RM, Espinosa MC, and Cobb LC.

(1994). Blinded comparison of cranial CT and MR in closed head injury evaluation.

AJNR Am J Neuroradiol 15, 351356.

26. Gentry LR, Godersky JC, Thompson B, and Dunn VD. (1988). Prospective

comparative study of intermediate-field MR and CT in the evaluation of closed head

trauma. AJR 150, 673682

27. Adams JH, and Duchen LW. (1992). eds. Greenfields neuropathology, 5th edn.

London: Edward Arnold: 106152.

28. Lee TT, Galarza M, and Villanuera PA. (1998). Diffuse axonal injury is not

associated with elevated intracranial pressure. Acta neurochir (wien) 140, 41-46.

29. Lagares A1, Ramos A, Alday R, Ballenilla F, Prez-Nuez A, Arrese I, Aln

JF, Pascual B, Kaen A, Gmez PA, and Lobato RD. (2006). Magnetic resonance in

moderate and severe head injury: comparative study of CT and MR findings.

Characteristics related to the presence and location of diffuse axonal injury in MR.

Neurocirugia 17, 105-118.

30. Atzema C, Mower WR, Hoffman JR, Holmes JF, Killian AJ, and Wolfson AB

(2006): National Emergency X-Radiography Utilization Study (NEXUS) II Group:

Prevalence and prognosis of traumatic intraventricular hemorrhage in patients with

Page 26 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

27

blunt head trauma. J Trauma 60: 1010-1017

31. Lee JP, Lui TN, and Chang CN (1991): Acute post-traumatic intraventricular

hemorrhage analysis of 25 patients with emphasis on final outcome. Acta Neurol

Scand 84:85-90,

32. Jayakumar PN, Sastry Kolluri VR, Basavakumar DG, Arya BYT, and Das BS

(1990). Prognosis in traumatic intraventricular haemorrhage. Acta Neurochir

106:4851.

33. Sato M, Tanaka S, Kohama A, and Fujii C (1987). Traumatic intraventricular

haemorrhage. Acta Neurochir 88:95103.

Figure legends

Page 27 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

28

Figure 1: Relationship between the IVH score on initial CT and DAI stages on MRI,

using nonparametric comparisons with control using Steel method. Compared to DAI

stage 0, the IVH score were significantly higher in DAI stage 2 and 3. The difference of

IVH score was not statistically significant between DAI stage 0 and stage 1.

Page 28 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

29

Figure 2: Relationship between IVH score on initial CT and the stage of DAI on

subsequent MRI. A 56 year-old male who was hit by a car while riding a bicycle shows

on initial CT performed about 30 minutes after onset (A), IVH in the posterior horn of

bilateral lateral ventricle (arrows), as the only post-traumatic lesion on CT. The patient

was rated IVH score 2. The subsequent MRI performed 3 hours later reveals

hypointense foci consistent with DAI in the right frontal lobe (arrow in B), in the corpus

callosum (arrow in C), and in the midbrain (arrows in D & E). The DWI (F) also shows

a non- hemorrhagic DAI lesion in the left frontal lobe (arrow). IVH in the posterior horn

of bilateral lateral ventricle are confirmed on T2*WI (C). The patient was diagnosed

with DAI stage 3 showing brain stem involvement.

Page 29 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

30

Page 30 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

31

Table 1 General characteristic of population

Variable Effectives

(%)

DAI

Group

n=48

Non-DAI

Group

n= 92

Univariate

analysis

Multivariate

analysis

Age* 41.7 (22.7) 41.3 41.9 (23.9) 0.9632 0.3885

Sex 0.0169* 0.0329*

male 102 (72.9) 41 (85.4) 61 (66)

female 38 (27.1) 7 (14.6) 31 (34)

GCS 0.0004* 0.0040*

Mild 89 (63.6) 21 (43.7) 68 (73.9)

Moderate 27 (19.3) 13 (27.1) 14 (15.2)

Severe 16 (11.4) 9 (18.7) 7 (7.6)

Missing 8 (5.7) 5 (10.5) 3 (3.3)

MOI 0.1385 0.5359

Traffic accident 94 (67.1) 37 (77.1) 57 (62)

fall 43 (30.7) 10 (20.8) 33 (35.9)

others 3 (2.1) 1 (2.1) 2 (2.1)

Time accident to MRI# 7.6 (8) days 7.8 (7.8) 7.5 (8.2) 0.5699 0.7754

*, mean (SD) in year; #, mean (SD) in days; GCS, Glasgow Coma Scale; MOI, Mean of

injury; MRI, Magnetic resonance imaging

Page 31 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

32

Table 2: Relationship between initial CT findings and presence of DAI on

subsequent MRI on logistic regression, in which SAH and IVH were treated as two

variables

Univariate analysis Multivariate analysis

CT items No. (%) Odd ratio 95% CI

(lowerupper

quintile)

P Adjusted

Odd ratio

95% CI

(lowerupper

quintile)

P

Basal cistern status 0.1653 0.1887

Normal 129 (92.1) 1.4 0.3-6.6 0.6713 1.5 0-21.7 0.7853

compressed 7 (5) 1.2 e-7 0-1.2 0.0680 1e-7 0-1.3 0.0678

absent 4 (2.9) 1.7 e-7 0-1.2 0.0665 1.5 0-3.1 0.1426

Positive midline shift 10 (7.1) 1.2 0.3-5.9 0.7647 1.4 0-25 0.8350

Positive EDH 17 (12.1) 2.7 0.8-12.1 0.1055 4 1-23 0.0534

Positive SAH 60 (42.9) 2 0.9-4.1 0.0512 1 0.4-2.7 0.9689

Positive IVH 27 (19.3) 3.7 1.6-9 0.0030* 4.2 1.3-14.3 0.0139*

Hemorrhagic mass 0.3509 0.5361

Absent 69 (49.3) 1.7 0.8-3.4 0.1649 1.7 0.7-4 0.2651

< 25 mL 67 (47.9) 0.5 0-4 0.5313 0.8 0-9.4 0.8607

>25 mL 4 (2.8) 0.8 0-6.8 0.8623 1.3 0-16.7 0.8330

EDH, epidural hematoma; SAH, subarachnoid hemorrhage; IVH, intraventricular

hemorrhage; CI, Confident interval

Page 32 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

33

Annexed table: Relationship between initial CT findings and presence of DAI on

subsequent MRI on logistic regression, in which IVH and SAH were combined and

treated as one variable of IVH/SAH

Univariate analysis Multivariate analysis

CT items No. (%) Odd

ratios

95% CI

(lowerupper

quintile)

P Adjusted

Odd ratio

95% CI

(lowerupper

quintile)

P

Basal cistern status 0.1653 0.1244

Normal 129 (92.1) 1.4 0.3-6.6 0.6713 0.9 0-13 0.9493

compressed 7 (5) 1.2 e-7 0-1.2 0.0680 7.9 e-8 0-1 0.0521

absent 4 (2.9) 1.7 e-7 0-1.2 0.0665 7.2e-8 0-1.4 0.0701

Positive midline shift 10 (7.1) 1.2 0.3-5.9 0.7647 2.2 0.1-65.2 0.5591

Positive EDH 17 (12.1) 2.7 0.8-12.1 0.1055 3.8 0.9-21.2 0.0578

Positive IVH/SAH 61 (43.6) 2.2 1.1-4.5 0.0290* 2 0.9-4.5 0.1120

Hemorrhagic mass 0.3509 0.5470

Absent 69

(49.3)

1.7 0.8-3.4 0.1649 1.6 0.7-3.8 0.2800

< 25 mL 67

(47.9)

0.5 0-4 0.5313 0.7 0-8 0.7662

>25 mL 4 (2.8) 0.8 0-6.8 0.8623 1 0-14 0.9416

EDH, epidural hematoma; SAH, subarachnoid hemorrhage; IVH, intraventricular

hemorrhage; CI, Confident interval

Positive IVH/SAH representing patients with either IVH or SAH or both.

Page 33 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.

34

Relationship between initial CT findings and presence of DAI on subsequent MRI on

logistic regression, in which IVH and SAH were combined and treated as one variable of

IVH/SAH

Page 34 of 34

Jour

nal o

f Neu

rotra

uma

Intra

vent

ricul

ar H

emor

rhag

e on

Initi

al C

ompu

ted

Tom

ogra

phy

as M

arke

r of D

iffus

e A

xona

l Inju

ry aft

er Tr

auma

tic B

rain I

njury

(doi: 1

0.108

9/neu

.2014

.3453

)Th

is ar

ticle

has

bee

n pe

er-re

view

ed a

nd a

ccep

ted

for p

ublic

atio

n, b

ut h

as y

et to

und

ergo

cop

yedi

ting

and

proo

f cor

rect

ion.

The

fina

l pub

lishe

d ve

rsio

n m

ay d

iffer

from

this

proo

f.