1245

Computed Tomographic-Angiographic
Findings Within the First Five Hours of
Cerebral Infarction
S.H. Horowitz, MD; J.L. Zito, MD; R. Donnarumma, RN, MA;
M. Patel, MD; and J. Alvir, DrPH

Background and Purpose: Modern management of acute stroke necessitates early diagnosis.
To this end, we sought to delineate the radiographic features of focal hemispheric infarction
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017

within 5 hours of ictus.

Methods: Fifty patients, ages 54-79, with ischemic strokes productive of at least hemiparesis
underwent computed tomographic scanning and cerebral angiography (n=38) or carotid ultra-
sound (n=12). Radiographic lesions were characterized for location, size, and pathophysiology.
Results: Acute abnormalities, hypodensity, and mass effect were seen in 56% of scans and
confirmed on a second scan 5-7 days later. Intracranial angiographic abnormalities occurred
in 61% of patients: arterial occlusions in 45% and delayed arterial filling in 16%. Hemorrhagic
infarctions occurred in 26% of second scans and were associated with mass effect (100%) and
arterial occlusions (89%). Infarcts with hemorrhagic transformation were larger on both scans
than those without (p=0.001). Of four patients with infarctions in watershed territories on the
scans, two had middle cerebral artery occlusions on angiography, thereby questioning the
specificity of such scan lesions to low-flow states.
Conclusions: We conclude that cerebral infarctions are often visible on early scans, but their
locations may not be etiologically determinative. The infarcts associated with intracranial
arterial occlusions (45%) were of thromboembolic origin, but, given current controversies as to
the pathophysiology of lacunar and watershed infarctions, we cannot ascertain the etiology in
the remainder. These findings are relevant to the new stroke therapies that require adminis-
tration in the first hours after infarction. (Stroke 1991^2:1245-1253)

M odern approaches to patients with cerebral

infarction emphasize early diagnosis and
management.1'2 Sophisticated computed
tomographic (CT) scanners, angiography, and Dopp-
ler/ultrasound of the carotid and intracranial circu-
lations increase the possibility of radiological confir-
mation of a suspected cerebral infarct within the first
few hours after ictus and aid in its localization and
pathophysiological characterization. However, delays
in implementation limit the value of angiography and
From the Departments of Neurology (S.H.H., R.D.), Radiology
(Neuroradiology) (J.L.Z., M.P.) and Psychiatry (Research) (J.A.),
Long Island Jewish Medical Center, the Long Island campus for
the Albert Einstein College of Medicine, New Hyde Park, N.Y.
Presented in part at the 16th International Joint Conference on
Stroke and Cerebral Circulation, San Francisco, Calif., February
1991.
Supported in part by Fidia Pharmaceuticals, SpA, Abano
Terme, Italy.
Address for correspondence: Dr. Steven H. Horowitz, Depart-
ment of Neurology, Long Island Jewish Medical Center, New
Hyde Park, NY 11042.
Received February 19, 1991; accepted June 11, 1991.
ultrasound due to frequent early recanalization of
thrombi3 or clot lysis.4 In order to delineate early
focal ischemic events and correlate the findings of
current clinical, CT, and angiographic or carotid
ultrasound techniques, we studied 50 patients within
5 hours of onset of infarction. The radiographic
findings are reported here and the clinical-radio-
graphic correlations will be reported (S. Horowitz, J.
Zito, R. Donnarumma, M. Patel, J. Alvir, unpub-
lished observations).
Subjects and Methods
Between November 1, 1987, and January 1, 1990,
50 consecutive patients with acute onset of focal
cerebral dysfunction, who gave informed consent,
presented with hemispheric infarction (right=20,
left=30) of 3 hours' or less duration, accurate timing
of onset, and motor deficit (hemiparesis). The exclu-
sion criteria consisted of inaccurate determination of
onset (patients who awakened with deficits were
excluded); previously known cerebral disease; non-
ischemic etiology, that is, hemorrhage, tumor, or
 1246 Stroke Vol 22, No 10 October 1991
infection; other life-threatening illnesses; coma; clin-
ical improvement within the first 5 hours, considered
to be transient ischemic attacks; focal cerebral dys-
function without hemiparesis; and brainstem involve-
ment. There were 33 males and 17 females, ages
54-79 (mean 67.9) years.
Noncontrast CT scans were performed on all pa-
tients using a GE 9800 scanner (Milwaukee, Wis.) with
10-mm cuts displayed on a 512x512 matrix (2-second
scan time) from 45 minutes to 4 hours after onset of
ictus, and were repeated at 5-7 days. The scans were
visually assessed for areas of decreased attenuation
(hypodensity); mass effect, as evidenced by sulcal ef-
facement, ventricular compression, or contralateral
shift of midline structures; hemorrhagic infarction, de-
fined as small, multifocal hyperdensities within a larger
hypodense area; localization of lesions to the vascular
territories of the three major cerebral arteries and
branches5-6; obscuration of the lentiform nucleus7'8;
hyperdensity of the middle cerebral artery (MCA)9-11;
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017
and antecedent lesions indicative of previously silent
infarcts. Lesion size was estimated as small (<Vi
lobe=l), medium (V2-I lobe=2), and large (>1
lobe=3). Only lesions situated in areas anatomically
compatible with the acute clinical deficits were consid-
ered. Infarcts were characterized as "territorial" if they
occurred in a discrete intracranial arterial distribution
and "watershed" if they occurred between two arterial
territories.612 The angiographic results contributed to
this differentiation.
Cerebral angiography was performed in 38 patients
immediately after the initial scan and completed within
5 hours of ictus. The clinically appropriate extracranial
carotid and intracranial vasculature was studied via
transfemoral catheterization with a 15% solution of
meglumine iothalamate (Conray R). In 12 patients,
both carotid arterial systems were studied. Verte-
brobasilar studies were not performed.
The extracranial arteries were assessed for athero-
sclerotic plaques and ulcers, kinking, intraluminal
filling defects, turbulence of blood flow, and stenosis.
Stenosis was graded on a scale of 0-4 with grade 1 as
<30%; grade 2, 31-74%; grade 3, 75-99%; and
grade 4, total occlusion. Grades 3 and 4 were consid-
ered hemodynamically significant.
The intracranial arteries were designated accord-
ing to the classification of Salamon and Huang13 and
assessed for the presence of stenosis or intraluminal
filling defects, occlusion, delayed flow, or collateral
circulation. Occlusions or stenoses of the MCA were
classified as follows: type I, proximal to the origin of
the lenticulostriate branches; type II, distal to the
lenticulostriate origins; type III, distal to the divi-
sional bifurcation involving either the superior (type
His) or inferior (type Illi) divisions; or type IV,
individual branch involvement.8'14
Intracranial arterial flow abnormalities were char-
acterized as complete occlusion, occlusion with ret-
rograde filling via a patent collateral circulation, and
slow filling with or without arterial stenosis. Slow or
delayed filling was in evidence when there was per-
sistence of contrast material in an artery into the
venous phase (>5.5 seconds after the onset of vascu-
lar opacification).
Carotid ultrasound was performed on the remain-
ing 12 patients within 5 hours of ictus. A similar
grading scale assessed carotid stenosis.
The data were analyzed with the \2 test for differ-
ences in proportions and using the t test for differ-
ences in means.
We wish to note a methodological caveat in our data
analysis. These patients participated in a randomized
double-blind study of GM1 ganglioside (Fidia, SpA,
Abano Terme, Italy) in patients with early infarcts.
The clinical examination, first CT scan, and angiogram
were performed before drug administration. We in-
cluded the second scan for purposes of corroboration
and further definition of the initial findings. In so
doing, we made the assumption that neither GM1 nor
placebo affected the second CT scan in a manner or to
a degree that would compromise these purposes. We
believe that the findings of this scan reflect the patho-
physiological changes that underlie the radiographic
features of an infarct more accurately than do the
findings of the first scan, especially in the absence of
new infarctions or anticoagulant therapy in any pa-
tient in the 5-7-day period between the initial ictus
and the second scan. To exclude the second scan is
unsatisfactory in that it precludes a discussion of
hemorrhagic infarction and limits our ability to make
accurate CT-angiographic correlations in this paper
and clinical-radiographic correlations elsewhere (S.
Horowitz, J. Zito, R. Donnarumma, M. Patel, J. Alvir,
unpublished observations). At this time, the random-
ized code has not been broken and we are unaware of
who received GM1 or placebo.
Results
Evidence of acute infarction occurred on the first
CT scan in 28 patients (56%) and on the second scan
in 37 patients (74%). In 11 patients, the first scan was
normal and the second abnormal; in 11, both scans
were normal; and in two, the first scan was abnormal
and the second normal.
Table 1 shows the radiographic features, size, and
location of lesions on both CT scans. Typical lacunar
infarctions (circumscribed hypodensities in the basal
ganglia-internal capsular area <1.5 cm in diameter)
occurred in two patients. In another, a lacunar infarc-
tion was seen along with hypodensity and mass effect in
the MCA cortical distribution. Infarct size was the
same on both scans in 16 patients. In five patients,
lesion size increased from small to medium; in two,
from small to large; and in three, from medium to large.
Of 11 patients with normal first scans and abnormal
second scans, nine developed small lesions, one a
medium lesion, and one a large lesion.
Mass effect was present on the first CT scan in 19
patients and on the second scan in 24 patients. On
the first scan, mass effect occurred without hypoden-
sity in four patients (8%); all had sulcal effacement
(Figure 1). The lesions were small in two patients and
 Horowitz et al Radiographic Features of Early Stroke 1247

TABLE 1. Summary of Computed Tomograhic Scan Results in 50 Patients

First CTscan Second CTscan
n % n %
Radiographic features
No acute abnormalities 22 44 13 26
Hypodensity alone 7 14 12 24
Hypodensity with mass effect 15 30 11 22
Mass effect alone 4 8 0 0
Hemorrhagic infarction with mass effect 0 0 13 26
Lentiform nucleus obscuration alone 2 4 1 2
Antecedent lesions 17 34

Lesion size
Small 13 26 14 28
Medium 12 24 14 28
Large 3 6 9 18

28 37
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017

Localization of infarcts by specific intracranial

artery distribution
Anterior cerebral artery 0 1
Middle cerebral artery 22 29
Cortical 15 16
Mixed cortical-subcortical 6 9
Subcortical 1 4'
Posterior cerebral artery 1 1
Lacunar infarcts 2 2
Watershed infarcts 3 4

28 37
CT, computed tomographic.
'One in association with an anterior choroidal artery infarct.

medium in two patients. Mass effect was more exten-
sive on the second CT scan and was always present
when hemorrhagic infarction occurred. In seven pa-
tients without mass effect on the first scan, sulcal
effacement with (six patients) and without (one pa-
tient) ventricular compression was seen on the sec-
ond scan. In 10 patients in whom sulcal effacement
was the only evidence of mass effect on the first scan,
additional ventricular compression was seen on the
second scan. A shift of midline structures was not
present on any first scan, but was seen, together with
sulcal effacement and ventricular compression, on
the second scan in nine patients, seven of whom had
lesser evidence of mass effect on the first scan
(Figure 1). Two patients with sulcal effacement on
the first scan did not demonstrate evidence of mass
effect on the second scan.
Mass effect was directly related to infarct size. On
the first scan, five of 13 (38%) small, 11 of 12 (92%)
medium, and all three large infarcts exhibited mass
effect O2=36.94, df=3, /><0.0001). Mass effect on
the first scan heralded larger lesions on the second
scan. None of the 14 small lesions seen on the second
scan had evidence of mass effect on the first scan,
whereas 11 of 14 (79%) medium and 7 of 9 (78%)
large lesions did so (*2=29.45, df=3,/?<0.0001).
The findings of the internal carotid artery (ICA)
on angiogram and ultrasound are documented in
Table 2. Of the 12 patients with angiographic hemo-
dynamically significant ipsilateral ICA stenoses, six
had complete MCA occlusions, three had MCA
occlusions with partial retrograde filling, two (grade
3) had delayed flow, and one (grade 3) had normal
intracranial flow. Tandem lesions of an ICA plaque
and an intracranial arterial occlusion occurred in
three patients with lesser stenoses (^grade 2) on
angiography. Common carotid or external carotid
artery abnormalities were rare (<5%).
Nine of 28 (32%) patients with positive first CT
scans and eight of 22 (36%) patients with negative
first scans had grade 3 or 4 ICA stenosis. Thirteen of
37 patients (35%) with positive second scans and four
of 13 patients (31%) with negative second scans had
such ICA stenosis (p=NS).
Of the 38 patients undergoing angiography, 20
(53%) had abnormal first CT scans and 23 (61%) had
abnormal intracranial angiograms (Table 3). There
was a significant association between the presence of
acute CT changes and the occurrence of occlusive or
hemodynamic intracranial angiographic abnormali-
ties (x =6.10, df=2,/j=0.047) and between the size
 1248 Stroke Vol 22, No 10 October 1991

FIGURE 1. Initial (left panels)

and repeat (right panels) com-
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017

puted tomographic scans of

three patients (A, B, and C) with
acute infarctions, demonstrating
sulcal effacement (black ar-
rows), ventricular compression
(curved white arrow), and hypo-
density (white arrows). Left
hemisphere is on reader's right.
Time from onset of ictus to ini-
tial scan: patient A, 2 hours and
15 minutes; patient B, 3 hours
and 15 minutes; patient C, 1
hour and 30 minutes. In patient
B, mass effect almost completely
obliterated the Sylvian fissure on
both scans (also seen on other
sections). In patient C, a contra-
lateral shift of the septum pellu-
cidum is visible.

of the infarct and these same arterial abnormalities cranial angiographic abnormalities (^-2=7.29, df=2,
(*2=17.83, df=6,p=0.007). /?=0.O27). Of importance is the association of type I
Acute changes on the second scan occurred in 26 and type II angiographic findings with larger CT
patients (68%) and also correlated with the intra- lesions (*2=28.15, df=6, />=0.0001). In the 12 pa-
 Horowitz et al Radiographic Features of Early Stroke 1249

TABLE 2. Hemodynamlc Findings of the Internal Carotid Artery on Angiogram and Ultrasound
Degree of stenosis Angiogram Ultrasound Totals
Ipsilateral to side of lesion
None or < grade 2 26 (3) 7(1) 33 (4)
Grade 3 ' 5t(3) 3(2) 8 (5)
Grade 4 ' 7 (2) 2 9 (2)
Total 38 (8) 12(3) 50(11)

Contralateral to side of lesion

None or <grade 2 9 (1) 11(2) 20 (3)
Grade 3* 2t(l) 1 3 (1)
Grade 4* 1 0 1

Total 12 (2) 12(2) 24 (4)

Numbers in parentheses indicate arteries with atherosclerotic plaque formation. In the case of grade 4 occlusions,
plaques were seen proximal to occlusions.
'Hemodynamically significant.
tOne patient had bilateral grade 3 stenosis.
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017

tients with normal second CT scans, one had a type second CT scans. The two patients with normal
IV occlusion, three had slow-flow states, and the second scans had type IV occlusions. Patients with
remaining eight (67%) had normal angiograms. flow abnormalities as the sole angiographic finding
Intracranial occlusive disease was associated with a were less likely to have acute CT changes (two of six
high frequency of CT lesions (Table 3). Thirteen of on the first scan, three of six on the second scan).
the 17 patients (76%) with MCA occlusions had Fifteen patients (39%) had normal intracranial vas-
abnormal first CT scans and 15 (88%) had abnormal culature: in 10 the first CT scan was normal, two had
lacunar infarctions, one had lentiform nucleus obscu-
ration, and two had infarcts in the MCA distribution.
TABLE 3. Correlation of First Computed Tomographic Scan
Results With Intracranial Angiogram Results (n=38) Two patients with normal first scans developed MCA
infarcts visible on the second scan, bringing to four
First computed the number of patients with normal angiograms and
tomographic scan Intracranial angiogram
territorial infarcts on CT.
Finding n % Finding n %
Abnormal 20 53 Arterial occlusions 13 65
In four patients (8%), infarcts were localized ex-
clusively to watershed or border zone areas, three on
Type I 6*
both CT scans and one only on the second scan.
Type II 1
Three were cortical temporoparieto-occipital and
Typenis 5t one was subcortical. In one patient with a temporo-
Type mi 0 parieto-occipital infarct, angiography revealed a type
Type IV 1 Illi MCA occlusion. The patient with a subcortical
Slow flow 2 10 infarct had a type I MCA occlusion. In the two
Type Ills 2 remaining patients, one had a grade IV ICA occlu-
Normal 5 25 sion on ultrasound and one had a grade I ICA
stenosis with type Illi MCA slow flow on angiogra-
Normal 18 47 Arterial occlusions 4 22 phy. Thus, only the latter patient, without evidence of
Type I 2
intracranial occlusive disease, had a true watershed
infarct. The patient with grade IV ICA occlusion on
Type II 0
carotid ultrasound may have had a true watershed
Types Ills and Illi 0 infarct, but the status of his intracranial vasculature
TypelV 2 is unknown. The other two patients had definite
Slow flow 4 22 thromboembolic disease that mimicked watershed
Type I 1 infarction on CT scan.
Type II 0 Hemorrhagic infarction was not visible on any first
Type Ills It scan, but was present on second scans in 13 patients
Type IV 2 (26%), accompanied by mass effect (no confluent
Normal 10 56 hematomas were seen). First-scan infarcts were evi-
dent in 12 patients (92%) who developed hemor-
'Anterior cerebral artery (ACA) occlusions in two, ACA slow rhagic infarction, as compared with 16 patients
flow in one.
tPartial type Illi occlusion in one, collateral flow in two. (43%) in whom hemorrhagic infarction did not
JWith collateral flow. evolve (/J=0.002, t test). Visualized lesions were
 1250 Stroke Vol 22, No 10 October 1991
larger on both scans in patients demonstrating hem-
orrhagic infarction (first scan, p=0.001; second scan,
p=0.001; t tests). Of patients without visible infarcts
on the first scan, only one of 22 (5%) developed
hemorrhagic infarction, whereas five of 13 patients
(38%) with small infarcts, four of 12 (33%) with
medium infarcts, and three of three (100%) with
large infarcts on the first scan developed hemor-
rhagic infarction subsequently. On the second scan,
one of 14 patients (7%) with small infarcts, five of 14
(36%) with medium infarcts, and seven of nine
(78%) with large infarcts exhibited hemorrhagic in-
farction. Nine patients with hemorrhagic infarction
underwent angiography; eight (89%) demonstrated
occlusive arterial lesions, as compared with nine of
the remaining 29 patients (32%) without hemor-
rhagic infarction (p=0.013, t test).
Hyperdensity of the MCA was seen on the first CT
scan in seven patients and in one on the second scan
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017
as well. In three cases, it occurred on the affected
side; in two, it was seen on the unaffected side on
both scans; and in two, it was present bilaterally on
both scans. In an eighth patient, bilateral hyperdense
MCAs were seen only on the second scan. In four
patients, calcification of the basilar artery was visible,
implying that the hyperdense MCAs were due to
atherosclerosis. Two patients (4%) had hyperdense
MCAs only on the first scan, only on the appropriate
side, and without basilar artery calcification, thereby
suggesting a thrombus or embolus within the artery.
Angiography was performed on five of eight pa-
tients demonstrating hyperdense MCAs on CT scan.
In the two with hyperdense MCAs on the affected
side only on the first scan, there was a complete ICA
occlusion (grade 4) in one and a type II MCA
occlusion in the other. In one patient with a hyper-
dense MCA on the unaffected side and in two
patients with bilateral hyperdense MCAs, the ICA
and main MCA trunk on the affected side were
normal: single MCA branch occlusions were seen.
Thus, there was corroborative angiographic evidence
of thromboembolic disease of the proximal MCA
only in the two patients with hyperdense MCAs on
the first scan.
Lentiform nucleus obscuration occurred in nine
patients on the first CT scan, but was the sole
abnormality in only two (4%). In one, it was the sole
abnormality on the second CT scan. In the remaining
seven patients, it was seen in association with hypo-
density and mass effect in the MCA distribution.
Discussion
Despite the high incidence of stroke, the number
of studies documenting its clinical and radiographic
features within the first few hours are relatively
scant. Some recent reports have noted slight hypo-
densities and mass effect on plain CT scans within 6
hours,2-7-11-15-16 but the general consensus is that
ischemic changes are usually not evident on a scan
until 24-48 hours after infarction.16 Only one
study8 has reported angiographic-CT correlations,
and only one2 has compared the radiographic find-
ings with the clinical picture within the first 6 hours.
Our results indicate that visible evidence of infarc-
tion on plain CT scans performed within the first 5
hours in patients with ischemic hemispheric strokes is
frequent (56%). Mass effect and hypodensity, alone
and together, are the major features seen in that time
frame. As expected, these changes are not as well
defined as those seen on the later scan, but they were
clearly distinguishable from the normal and localiz-
able (Figure 1).
Multiple experimental studies have documented
the pathophysiology of focal cerebral ischemia.
Within 1 hour after ischemia, a 3% increase in total
water occurs in the ischemic area due to vascular
dysautoregulation17 and development of intracellular
(cytotoxic) edema.1819 Extracellular (vasogenic) fluid
accumulates, in part dependent on regional perfu-
sion,19 presumably from collateral flow. Analysis of
the biomechanical properties of ischemic tissue indi-
cates the measurable presence of slight, but signifi-
cant, edema 1 hour after occlusion.20
The early CT scan evidence of cerebral infarction
is dependent on this accumulation of edematous
fluid. Increases in water content result in decreases
in brain tissue specific gravity, and CT attenuation is
linearly proportional to specific gravity.21 Changes of
2.5-2.6 Hounsfield units occur for every 1% change
in water content.22 However, visualization of an
infarct on CT is more than hypodensity associated
with specific gravity changes. In four of our first CT
scans (14% of positive first scans), mass effect is seen
alone, and in other scans the hypodensities are
almost always mild and contiguous with areas of mass
effect in which hypodensity is absent. In these situa-
tions, sulcal effacement and ventricular compression
reveal the presence of an infarct without, or in
association with, mild hypodensity. Sulcal effacement
and ventricular compression are caused by edema-
tous changes that do not alter specific gravity enough
to be visible as a hypodensity, but which are percep-
tible within several hours of ictus.
Tomura et al7 reported obscuration of the lenti-
form nucleus in 23 of 25 patients with embolic
infarctions on CT scans performed within 6 hours of
ictus. Lentiform nucleus obscuration occurred alone
in eight (32%), in association with slight hypodensity
in 11, and with slight hypodensity and sulcal efface-
ment in four. Bozzao et al8 found abnormalities in 25
of 36 plain CT scans within 4 hours. Lentiform
changes occurred alone in 14 (39%) and in associa-
tion with cortical hypodensity in five. Cortical hypo-
density or sulcal effacement without lentiform
changes was seen in the other six.
Lentiform nucleus obscuration alone was seen in
two patients (4%) on the first scan, in one of whom it
also occurred alone on the second scan (suggesting
that it was an artifact or not an acute change). This
finding has been attributed to the early development
of cytotoxic edema in the basal ganglia region,7 but
the presence of gross mass effect in conjunction with
 Horowitz et al Radiographic Features of Early Stroke
lentiform nucleus obscuration in seven of nine pa-
tients indicates that vasogenic edema is also present
within the first few hours and both forms of edema
proceed pan passu. Our results suggest that within
the first few hours after infarction, lentiform changes
do not occur without hypodensity or mass effect as
often as previously reported.7-8
Pressman et al9 found increased density in the
MCA on CT scans performed within 24 hours of ictus
in 15 patients. Schuierer and Huk10 reported unilat-
eral hyperdense MCAs in four patients within 6
hours. Tomsick et al11 reported its presence in seven
of 25 patients (28%) undergoing plain CT scan within
90 minutes of ictus. No other findings were noted.
In our series, hyperdense MCA was not a valuable
sign. While probably indicative of a thrombus or
embolus within the artery9-11 in the two patients
(4%) with ipsilateral hyperdensities only on the first
scan and ICA or type II MCA occlusions on angiog-
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017
raphy, in the other six patients it most likely repre-
sents atherosclerosis with vascular calcification.
Hemorrhagic infarction, not present on any first
scan, was seen on the second scan in 26% of patients.
In agreement with previous reports,23-27 infarcts with
visible hemorrhagic components were significantly
larger than those without hemorrhagic components.
We, and others,25'27 found this size factor to be
present on early scans even before hemorrhagic
features were evident. However, the development of
hemorrhagic infarction is not solely confined to
larger lesions. Patients without infarcts (5%) and
38% with small infarcts on the initial scan did
develop hemorrhagic infarction subsequently.
Arterial occlusions were seen in all but one (89%)
of the patients with hemorrhagic infarction who
underwent angiography. This high incidence of oc-
clusive disease in hemorrhagic infarction is compat-
ible with either the Fisher-Adams28 model of in-
creased vascular permeability following arterial
reperfusion after clot lysis or migration, or the con-
cept of reperfusion to the ischemic area via lepto-
meningeal collateral circulation if occlusion of the
feeding artery persists.29 As we did not perform serial
angiography or transcranial Doppler sonography, we
could not assess either of these possibilities.
The occurrence of mass effect on all CT scans
demonstrating hemorrhagic infarction supports the
findings of Lodder24 and Hornig et al25 that edema
and mass effect are more prominent in hemorrhagic
than in nonhemorrhagic infarction. The predilection
for larger infarcts to develop hemorrhagic infarction
with mass effect is also compatible with the theory
that extensive edema causes small-vessel compres-
sion and blood-flow stasis. Subsequent decreases in
edema may allow reperfusion of vessels, which by
then have endothelial disruption, and diapedesis of
blood follows.25
As many as 10% of all cerebral infarcts are claimed
to be watershed lesions30; the diagnosis is based on
characteristic CT infarcts situated between the bor-
der zones of two or more arterial territories.6-12^1-32
1251
Angiographic-CT correlations have associated CT
lesions in watershed areas with severe carotid artery
disease12-30-33 and have implicated perfusion pres-
sure hemodynamics in the production of focal cere-
bral hypoperfusion of the distalmost territories of
branched arteries.31
In this study, four patients (8%) exhibited evi-
dence of acute infarctions in watershed territories on
CT scan. Angiography demonstrated occlusion of
arteries whose territories supply the infarcted ar-
eas5-6-13 in two of these patients, one with a superficial
temporoparieto-occipital cortical infarction (MCA
type Illi) and one with a subcortical infarction (MCA
type I). Angeloni et al34 described seven patients with
distal pial MCA branch occlusions and a CT pattern
of subcortical watershed infarctions, and Torvik30
reviewed cases of watershed infarcts caused by either
cholesterol35 or tumor36 emboli. Taken in conjunc-
tion with these reports, our results suggest that
infarctions in the watershed areas on CT are not
necessarily specific or diagnostic of hemodynamic
"low-flow" states and may occur in occlusive disease
of thromboembolic etiology.
Our incidence of arterial occlusions is lower than
the 10% ICA and 66% intracranial occlusions of
Fieschi et al2 in 80 patients and the 83% incidence of
arterial occlusive disease of Bozzao et al8 in 36
patients. Both reports studied acute ischemic stroke
within 6 hours. While we concur with Bozzao and
colleagues regarding an association between early
acute changes on CT scan and intracranial arterial
disease, our results differ in degree. Whereas they
found intracranial arterial occlusions in all patients
with early CT findings, we documented occlusions in
65% of such patients. Another 10% of our patients
had delayed flow through the superior division of the
MCA, and 25% had normal intracranial angiograms
(Table 3). Further, we and Bozzao et al found
intracranial arterial occlusions in 22% and 45%,
respectively, of patients with normal early CT scans.
We, in addition, found delayed flow in 22% of
patients with normal early scans.
The presence of acute CT scan changes correlated
with the degree of intracranial occlusive disease.
With occlusions of two or more branch arteries
(MCA types IIII), evidence of infarction was visible
on 86% of first scans (Table 3) and on all second
scans. Severity of occlusive disease was associated
with lesion size on both scans. Intracranial flow
abnormalities without intracranial arterial occlusions
were less productive of acute CT changes (33% on
the first scan, 50% on the second scan), and normal
intracranial arteries were even less frequently asso-
ciated with abnormal scans (25% on the first scan,
27% on the second scan). In addition, severe ICA
stenosis (grade 3) or occlusion (grade 4) were not
related to visible infarcts on either scan.
Analyses of cerebral infarctions through the 1980s
focused on three types with, in large part, distinct and
separable etiologies: lacunar infarcts in small-vessel
disease (microatheroma or lipohyalinosis) secondary
 1252 Stroke Vol 22, No 10 October 1991
to hypertension37-40; watershed infarcts in low-flow
states resulting from carotid disease12-30"34; and ter-
ritorial infarcts of thromboembolic origin.612 The CT
correlates have been established, thereby allowing
for radiographic discernment of each type of infarct.
However, the results of this and other very recent
studies may temper these concepts.
New and convincing reports raise doubts as to the
select or even predominant role of hypertension in
lacunar states,4142 find that carotid disease with low
flow43 and cerebral microembolic disease42 can also
produce lacunes, and question the very existence of a
definitive ischemic condition with the "lacune" ap-
pellation.42-44 Similarly, we and Angeloni et al34 find
vascular occlusive disease on angiography in the
presence of watershed infarcts on CT scanning,
thereby disputing the exclusivity of low flow in the
etiology of this condition.
Our finding that four patients had normal angio-
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017
grams despite visible infarcts in the MCA distribu-
tion on CT scan, while not excluding thromboembolic
territorial infarction, is of particular interest. A clot
or an embolus could produce an infarct and break up
before the angiogram, or microemboli could lodge in
cerebral arteries too small to be visible on routine
angiography, but normal angiography within 5 hours
postictus raises the possibility of other mechanisms
for these infarctions.
Consequently, although it would be convenient to
subdivide our patients according to these three eti-
ologies, attributing infarcts that were not clearly
lacunar (4%) or watershed (4%) to thromboembo-
lism (92%), we are convinced that thromboembolic
processes exist only in the 45% of patients with
intracranial occlusions. We are uncertain as to the
infarct etiologies in the remaining patients. These
concerns are relevant in view of recent advances in
thrombolytic therapy and suggest that angiography or
transcranial Doppler sonography is important in the
workup of the early stroke patient.
Acknowledgments
We greatly appreciate V. Chehebar, MD; R. Reife,
MD; K. Bergmann, MD; R. Baumann, MD; S. Jain,
MD; and T. Kwiatkowski, MD, and his staff in the
Emergency Department for assistance in patient
recruitment. We also thank R. Konigsberg, DO; H.
Orzach, MD; and M. Stern, MD, and the Depart-
ment of Radiology staff for help in the performance
of the cerebral angiograms.
References
1. Grotta JC: Current medical and surgical therapy for cere-
brovascular disease. N Engl J Med 1987;317:1505-1516
2. Fieschi C, Argentino C, Lenzi GL, Sacchetti ML, Toni D,
Bozzao L: Clinical and instrumental evaluation of patients
with ischemic stroke within the first six hours. / Neurol Sci
1989;91:311-322
3. Fieschi C, Bozzao L: Transient embolic occlusion of the
middle cerebral and internal carotid arteries in cerebral
apoplexy. / Neurol Neurosurg Psychiatry 1969;32:236-240
4. Dalai PM, Shah PM, Sheth SC, Deshpande CK; Cerebral
embolism: Angiographic observations on spontaneous clot
lysis. Lancet 1965;l:61-64
5. Damasio H: A computed tomographic guide to the identifica-
tion of cerebral vascular territories. Arch Neurol 1983;40:
138-142
6. Ringelstein EB, Koschorke S, Holling A, Thron A, Lambertz
H, Minale C: Computed tomographic patterns of proven
embolic brain infarctions. Ann Neurol 1989;26:759-765
7. Tomura N, Uemura K, Inugami A, Fujita H, Higano S,
Shishido F: Early CT findings in cerebral infarction: Obscura-
tion of the lentiform nucleus. Radiology 1988;168:463-467
8. Bozzao L, Bastianello S, Fantozzi LM, Angeloni U, Argentino
C, Fieschi C: Correlation of angiographic and sequential CT
findings in patients with evolving cerebral infarction. AJNR
1989;10:1215-1222
9. Pressman BD, Tourje EJ, Thompson JR: An early CT sign of
ischemic infarction: Increased density in a cerebral artery.
AJNR 1987;8:645-648
10. Schuierer G, Huk W: The unilateral hyperdense middle
cerebral artery: An early CT sign of embolism or thrombosis.
Neuroradiology 1988;30:120-122
11. Tomsick TA, Brott TG, Chambers AA, Fox AJ, Gaskill MF,
Lukin RR, Pleatman CW, Wiot JG, Bourekas E: Hyperdense
middle cerebral artery sign on CT: Efficacy in detecting middle
cerebral artery thrombosis. AJNR 1990;ll:473-477
12. Ringelstein EB, Zeumer H, Angelou D: The pathogenesis of
strokes from internal carotid artery occlusion: Diagnostic and
therapeutical implications. Stroke 1983;14:867-875
13. Salamon G, Huang YP: Radiologic Anatomy of the Brain.
Berlin, Springer-Verlag, 1976, chap 3, pp 47-109
14. Saito I, Segawa H, Shiokawa Y, Taniguchi M, Tsutsumi K:
Middle cerebral artery occlusion: Correlation of computed
tomography and angiography with clinical outcome. Stroke
1987;18:863-868
15. Inoue Y, Takemoto K, Miyamoto T, Yoshikawa N, Taniguchi
S, Saiwai S, Nishimura Y, Komatsu T: Sequential computed
tomography scans in acute cerebral infarction. Radiology 1980;
135:655-662
16. Wall SD, Brant-Zawadzki M, Jeffrey RB, Barnes B: High
frequency CT findings within 24 hours after cerebral infarc-
tion. AJR 1982;138:3O7-311
17. Garcia JH: Experimental ischemic stroke: A review. Stroke
1984;15:5-14
18. Bell BA, Symon L, Branston NM: CBF and time thresholds for
the formation of ischemic cerebral edema, and effect of
reperfusion in baboons. / Neurosurg 1985;62:31-41
19. Brant-Zawadski M, Weinstein P, Bartkowski H, Moseley M:
MR imaging and spectroscopy in clinical and experimental
cerebral ischemia: A review. AJNR 1987;8:39-48
20. Hatashita S, Hoff JT: Biomechanics of brain edema in acute
cerebral ischemia in cats. Stroke 1988;19:91-97
21. Rieth KG, Fujiwara K, DiChiro G, Klatzo I, Brooks RA,
Johnston GS, O'Connor CM, Mitchell LG: Serial measure-
ments of CT attenuation and specific gravity in experimental
cerebral edema. Radiology 1980;135:343-348
22. Unger E, Littlefield J, Gado M: Water content and water
structure in CT and MR signal changes: Possible influence in
detection of early stroke. AJNR 1988;9:687-691
23. Cerebral Embolism Study Group: Immediate anticoagulation
of embolic stroke: Brain hemorrhage and management
options. Stroke 1984;15:779-789
24. Lodder J: CT-detected hemorrhagic infarction: Relation with
size of the infarct, and the presence of midline shift. Acta
Neurol Scand 1984;70:329-335
25. Hornig CR, Dorndorf W, Agnoli AL: Hemorrhagic cerebral
infarction: A prospective study. Stroke 1986;17:179-185
26. Hart RG, Easton JD: Hemon-hagic infarcts. Stroke 1986;17:
586-589
27. Okada Y, Yamaguchi T, Minematsu K, Miyashita T, Sawada
T, Sadoshima S, Fujishima M, Omae T: Hemorrhagic trans-
formation in cerebral embolism. Stroke 1989;20:598-603
 Horowitz et al
28. Fisher CM, Adams RD: Observations on brain embolism with
special reference to the mechanism of cerebral infarction. /
Neuwpathol Exp Neuml 1951;10:92-94
29. Ogata J, Yutani C, Imakita M, Ishibashi-Ueda H, Saku Y,
Minematsu K, Sawada T, Yamaguchi T: Hemorrhagic infarct
of the brain without a reopening of the occluded arteries in
cardioembolic stroke. Stroke 1989;20:876-883
30. Torvik A: The pathogenesis of watershed infarcts in the brain.
Stroke 1984; 15:221-223
31. Bogousslavsky J, Regli F: Unilateral watershed cerebral in-
farcts. Neurology 1986^6:373-377
32. Bogousslavsky J, Regli F: Borderzone infarctions distal to
internal carotid artery occlusion: Prognostic implications. Ann
Neurol 1986;20:346-350
33. Rodda RA: The arterial patterns associated with internal
carotid disease and cerebral infarcts. Stroke 1986;17:69-75
34. Angeloni U, Bozzao L, Fantozzi L, Bastianello S, Kushner M,
Fieschi C: Internal borderzone infarction following acute
middle cerebral artery occlusion. Neurology 1990;40:
1196-1198
35. Beal MF, Williams RS, Richardson EP, Fisher CM: Choles-
terol embolism as a cause of transient ischemic attacks and
cerebral infarction. Neurology 1981;31:860-865
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017
Radiographic Features of Early Stroke 1253
36. Torvik A, Skullerud K: Watershed infarcts in the brain caused
by microemboli. Clin Neuwpathol 1982;l:99-105
37. Fisher CM: Lacunes: Small deep cerebral infarcts. Neurology
1965;15:774-784
38. Fisher CM: The arterial lesions underlying lacunes. Ada
Neuwpathol 1969;12:1-15
39. Caplan LR, Stein RW: Stroke: A Clinical Approach. Boston,
Butterworths, 1986, pp 167-177
40. Bamford JM, Warlow CP: Evolution and testing of the lacunar
hypothesis. Stroke 1988;19:1074-1082
41. Bamford JM, Wallow CP: Are hypertension or cardiac embo-
lism likely causes of lacunar infarction? Stroke 1990;21:
375-381
42. Millikan C, Futrell N: The fallacy of the lacune hypothesis.
Stroke 1990;21:1251-1257
43. Waterston JA, Brown MM, Butler P, Swash M: Small deep
cerebral infarcts associated with occlusive internal carotid
artery disease: A hemodynamic phenomenon? Arch Neurol
1990;47:953-957
44. Landau WM: Au clair de lacune: Holy, wholly, holey logic.
Neurology 1989;39:725-730
KEY WORDS angiography cerebral infarction tomography,
x-ray computed
 Computed tomographic-angiographic findings within the first five hours of cerebral
infarction.
S H Horowitz, J L Zito, R Donnarumma, M Patel and J Alvir

Stroke. 1991;22:1245-1253
doi: 10.1161/01.STR.22.10.1245
Downloaded from http://stroke.ahajournals.org/ by guest on September 16, 2017

Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 1991 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/22/10/1245

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click Request
Permissions in the middle column of the Web page under Services. Further information about this process is
available in the Permissions and Rights Question and Answer document.

Reprints: Information about reprints can be found online at:

http://www.lww.com/reprints

Subscriptions: Information about subscribing to Stroke is online at:

http://stroke.ahajournals.org//subscriptions/