Ronald
G. Quisling,
MD
Persistent Arteriovenous Radlosurgery:
#{149} Keith
Nidus
R. Peters,
MD
#{149} William
A. Friedman,
MD
T
treatment of arteriovenous malformations (AVMs) of the brain has evolved in recent years from primarily a surgical approach to one HE
that
includes
the
adjunctive
use
of
stereotactic radiosurgery and embolization therapy (1-7). This multimodality approach presents new demands for magnetic resonance (MR) imaging (8,9). Pretreatment MR imaging must now enable clarification of the morphologic aspects of the afferent and efferent AVM blood supply, identification of the exact location of the AVM matrix, and estimation of the flow characteristics across all compartments of the nidus. It is no longer sufficient to limit the role of MR imaging to defining the presence or absence of a brain AVM. The effects of radiosurgery are being evaluated in current clinical practice, but much is already known. The preliminary results of radiosurgical treatment suggest that nidus obliteralion occurs in pial and deep AVMs in 30%-40% of cases within 1 year and 75%-85% within 2 years (4,5,7). Similar results are being observed in proton beam, gamma knife, and now linear accelerator-Linac (3,4,7,10) treatments.
Posttreatment MR imaging objectives include assessment of radiationinduced brain changes and residual AVM nidus blood flow (transnidus flow). MR imaging can help monitor the size of such lesions in a noninvasive manner (11-19). However, estimation of residual transnidus blood flow requires the use of special tech-
terms:
Arteriovenous malformations, 17.75 #{149}Brain, MR studies, 17.1214 blood vessels, flow dynamics, #{149} Magnetic resonance (MR), vascular 17.75
cerebral,
Cerebral 17.1214 studies,
Radiology
1991;
180:785-791
niques,
I
From
the
Departments
of Radiology
K.R.P., R.P.T.) and Neurosurgery versity
©
to R.G.Q.
RSNA,
1991
Medical
Center,
Uni-
Shands Hospital, J.H. Miller Health Center, Box J-374, Gainesville, FL 32610. From the 1990 RSNA srientific assembly. Received January 16, 1991; revision requested March 4; revision received March 20; accepted April 1. Address reprint re-
quests
of Florida
(R.G.Q.,
(W.A.F.),
P. Tart,
MD
Blood Flow In Cerebral Malformation after Stereotactic MR Imaging Assessment’
Stereotactic radiosurgery has become a major force in the treatment of arterioveous malformations (AVMs) of the brain. After treatment, obliteration of flow through the malformation occurs in 75%-85% of cases within 2 years, assuming the entire AVM nidus can be encompassed by the radiation field. Because the follow-up period is relatively long, a noninvasive means to assess residual transnidus blood flow is desirable. The authors report favorable findings after a comparative analysis of 85 posttreatment magnetic resonance images and 27 follow-up cerebral arteriograms in 34 patients treated with stereotactic radiosurgery. The authors found that transnidus flow can be determined from apparent signal intensity differences between tandem two-dimensional gradient-recalled echo images obtained first without and then with gradient moment nulling (flow compensation), with empirically derived pulse parameters. This method provides a means to monitor the reduction in AVM matrix size and to assess the extent of persistent arteriovenous shunting (ie, blood flow) across the nidus. Index
#{149} Roger
many
of which
are
relatively
new (eg, phase shift imaging and two- and three-dimensional MR angiography) (12-14). The merits of these methods remain to be proved. By contrast; gradient moment nulling (flow compensation) techniques are widely
available
to enhance by
signal
and
have
intensity
velocity-dependent
secondary
to nonstationary
phase
been
especially
that
associated
stant-velocity
blood
with
flow
con-
(16,17).
To our knowledge, no previously published study has employed twodimensional gradient-recalled echo (GRE) acquisitions performed in tandem first without flow compensation and then, immediately following, with flow compensation (see Patients and Methods). These acquisitions are performed
to estimate
residual
nidus
blood flow in radiosurgically treated vascular brain malformations. The goals of this study were to establish whether such an AVM MR imaging protocol could accurately depict the presence of residual flow across an AVM nidus, enable assessment of the progressive
diminution
of transnidus
flow over time, and enable determination of when complete thrombosis has occurred. PATIENTS
AND
METHODS
This comparative review of MR imaging and angiography included 34 patients Selected from a larger pool of patients who
had undergone
radiosurgery
to treat
cere-
bral AVMs. All patients in this study had undergone an initial angiographic evaluation followed by stereotactic radiosurgery
with
a Linac scalpel
Conn).
During
had been
(Philips,
the
Shelton,
follow-up
studied
with
period,
they
our AVM MR im-
aging protocol (ie, tandem GRE studies with and without flow compensation). A total of 85 posttreatment MR images
and 27 follow-up provided ysis.
cerebral
the basis
Thirteen
arteriograms
for this
patients
had
correlative
anal-
undergone
se-
rial MR imaging but had not completed follow-up cerebral arteriography. Follow-up MR images were generally obtamed at 6-month intervals after treatment, unless acute symptoms intervened. The distribution of MR images among patients is shown in Figure 1. At least one posttreatment
tamed
AVM
MR
in all 34 patients;
image
was
two images
ob-
were
used
degraded shifts
nuclei,
Abbreviations: mation,
GRE
=
AVM = arteriovenous gradient-recalled echo.
malfor-
785
obtained
in 24;
three
images
tamed in 17; four images six; and five posttreatment
were
obtained
bution
of MR
follow-up
were
Magnification
ob-
were obtained MR images
in
in one. The temporal images
period,
obtained
which
24 months,
is presented
MR images
were
distri-
ranged
from
compared
treatment
and
with
the
initial
an-
protocol angiography
in-
with
The cluded
clinical (radiosurgical) preliminary cerebral
formed
treatment
the
basis
planning.
was
scheduled
1-year
intervals
sisted.
Angiography
for
radiosurgical
Follow-up
sity
angiogra-
nidus
was
flow
per-
terminated
once
obliteration was achieved or no additional therapeutic response was apparent. To date, 27 follow-up angiograms have been obtained in 21 of 34 patients. Several patients have undergone more than one study,
while
others
complete
the
temporal
distribution
presented
(13
angiographic
MR
have
yet
The is
imaging
Studies
also
followed
a sagaxial
GRE studies region of the approximately
were
were
lim-
AVM
(Magnetom
performed
a repetition
msec,
a section
angle
of
msec,
four
time
thickness
signals
or
time
averaged,
mm, a flip of 12-15
a 256
were
not
with this method (Fig of nidus volumes were pretreatment calculations
the area of the stereotactic frame used to prescribe
786
#{149} Radiology
in seven
immagni-
Direct
tandem
[i7-i9])
over
the
sig-
GRE
the
changed
GRE
loca-
flow)
x 256
included
fields).
4) of calconsistent (minus
radiosurgical the radiation
ther-
treatment field. The second MR phy
but
recognition
thera-
comparison
imaging-angiografocused
on
whether
I
20
1
2
or
4
NO. OF MR SCANS
Figure tamed obtained
1. per
Distribution
5
PER PATIENT
of MR images
patient. Eighty-five in 34 patients.
ob-
images
were
no apparent
This
was
used
30 ‘u-i
for subtle
it added
of persistent
in five, in
10
isointense
when
to account
overall
mini-
30
difference in signal intensity could be discerned. Analog subtraction can be performed with the non-flow-compensated mask.
nidus
19 patients,
peutic consequence in the posttreatment period, provided that nidus volume progressively declines and that no additional nidus appears in contiguous regions outside the original
without
to either
as a subtraction
matched
of the
for
Re-
from hy-
image
was noted
imaging
15%
for three.
mally overestimated nidus size and underestimated nidus size seven compared with angiography. Errors of this degree have little
evinoted
clearly hyperintense (on the GRE image with flow compensation) relative to the brain in any part of the nidus. Negative evidence (indicative of no residual nidus
blood
within
25%
obtained
Positive flow was
intensities (on
in
similar
image
three,
im-
increased
precisely
second
signal
pointense
changes,
in the nidus volume calculations on any of the pre- or posttreatment MR images or angiograpic studies. This is especially important because radiosurgery usually preserves the parent circulation. Thus, residual flow persisted in both the pial arteries and veins in and around the AVM, even after transnidus flow had been eliminated.
treatment
on
when
little
to the
transnidus
z
flow. (liz
RESULTS
of 140-iSO of 4-5
an echo
400_500,
apy
the
effects
Siemens, Imaging parameters of these studies
malformation
Results culation with the
between
only occasionally
with
matrix, a field of view of 16-20 mm, and an interleaved series. Flow compensation was applied in the section-selection and readout directions only. Measured residual nidus volumes were calculated from the regions demonstrating enhancement on the tandem GRE images. The afferent and efferent portions of the vascular
of MR
size
precise
nulling.
for
within
sults
SP A2.i;
Iselin, NJ) MR imager. for the GRE component included
difference
study
2.5
a i.5-T (Signa Performance Plus GE Medical Systems, Milwaukee)
a i.0-T
moment
10%
and
values.
be a practical the
two,
for the
on stereotactic
It would
flow compensation)
T2-weighted imaging, and serial, axial, and GRE sequences performed first without and then with flow compensation gra-
minutes.
based
with flow compensation. dence of residual blood
angiograms
specific protocol that included routine ittal and axial Ti-weighted imaging,
either 3.3.8;
value
to determine
gradient
tions
3.
dients applied. The ited spatially to the nidus; each required
region
time-of-flight signal intensity
to
follow-up. of the
in Figure
Interval
of 34)
measured
a
ages was our primary diagnostic concern. A GRE AVM image was considered abnormal (transnidus flow present) if the lowsignal-intensity nidus (due to fast-flow,
at approximately
if residual
to the
within
cerebral
by applying
nal intensity measurements within the nidus were not performed, in part because the increase in signal intensity within an AVM nidus was seldom uniform across all pixels defining the nidus and because the observable conspicuity of the signal inten-
(in all 34 patients) and dynamic computed tomography performed in a stereotactic radiosurgical frame. Once digitized, these studies
the
fication of each plane for each intracranial AVM. Qualitative evidence of persistent nidus flow was defined as an apparent increase in signal intensity between the gradient images obtained first without and then
pre-
cerebral
on
corrected
an extrapolated
possibility
giograms.
phy
was
measurements.
in a blinded
posttreatment
reduction
This
10 to
manner by two neuroradiologists (R.G.Q., K.P.), and quantitation of residual nidus flow was tabulated. MR images were subsequently
25%
midcranial
2. All
in Figure
reviewed
were
the
during
effects
angiograms
The tial
data
step,
were
nidus
analyzed. size
mi-
As an
correlation
was
0
performed to ensure that what was perceived as nidus by tandem (with and without flow compensation) GRE studies did, in fact, correlate with cerebral parison
angiographic of mdus
size
findings. at MR
3.6
6.12
MONThS
Figure tamed
ment. patients
Comimaging
2. per
POST
12-18
Distribution of MR images patient during follow-up
Eighty-five (PTS.).
images
18-24
TREATMENT
were
obtreat-
obtained
in 34
and angiography was performed in 19 of the 34 patients who had undergone
both
AVM
MR imaging
proximity
cerebral (all
weeks
or less
hours).
These
arteriography
in close
within
and studies
a period
most
of 4
within were
with
angiography
performed
for
U) 0 z 4
24
to determine whether measured nidus volumes determined at MR imaging matched those obtained with cutfilm angiography. These data are presented in Figure 5. Close size correlation was observed. Among the 18 patients, nidus volume at MR imaging was within 5% of that measured
30
and
temporal
eight,
20
0 0 z
Ia
I
10
0
(U
I 3.6 SIX
Figure
3.
Distribution
(ANGIOS) obtained seven angiograms tients.
Ih 6.12
MONTH
12-18
18.24
INTERVALS
of angiograms
after treatment. were obtained
Twentyin 34 pa-
September
1991
or false-negative flow seen with ual AVM nidus
diagnosis MR imaging
documentation of absence of active AVM residua; this is graphically portrayed in Figure 6. Thus, MR imaging enabled correct prediction when there was no residual nidus flow before confirmation with angiography in each of seven cases, as illustrated in
the second set included patients who had angiographic evidence of persistent transnidus flow (flow-positive
this AVM MR imaging protocol could be used to correctly determine if residual nidus flow had been obliterated. To assess the likelihood of a false-positive (ie, flow-positive MR image but flow-negative angiogram)
status).
In the first instance, comparative analysis was performed on 21 preangiographic MR images obtained in seven patients, who, at follow-up an-
(ie, no but resid-
seen with angiography), two data sets were studied. The first set included patients with angiographic confirmation of obliterated nidus (ie, flow-negative status), and
giography,
dus bosis).
had
flow
no
residual
(ie, complete
AVM
A progressive
Figure
were
transni-
decline
residual mdus volume was on the MR images preceding
7. No
false-positive
observed
in this
diagnoses
limited
set of
throm-
patients.
in
In the second part of this evaluation, 42 preangiographic MR images from 18 patients who exhibited angio-
observed final
graphic
evidence
of persistent
trans-
nidus flow were analyzed. All MR images in this data set demonstrated a direct correlation with angiograms, correctly suggesting persistent nidus flow when, in fact, it did exist. In none of the cases with angiographic evidence of residual transnidus flow did results at MR imaging suggest completed thrombosis (ie, no falsenegative diagnoses). To evaluate whether temporally distinct studies reflected the anticipated progressive diminution of residual transnidus flow and to determine whether nidus measurements were consistent between serial MR imaging examinations, the measured residual nidus volumes (after 3 months to 2 years) were compared in 78 MR images obtained from 26 patients. Patients for whom only one
b.
d.
C.
Figure 4. Measurement which measurements cluded in measurements. subtraction angiogram
with
direct
anteroposterior
image obtained be hypointense
e
parameters for MR imaging-angiography correlation. of length (L) and height (H) of AVM nidus are obtained. This corresponds to methods used to calculate nidus illustrates means by which measurement of nidus width
without relative
and
lateral
flow compensation. to adjacent brain
measurements Fast-flow, tissue. AVM
of height,
width,
time-of-flight, nidus, depending
and and
(a) Lateral
Major
carotid
arterial
subtraction
pedicles
angiogram
and
venous
illustrates
efferent
vessels
means
volume for radiosurgical purposes. (b) Anteroposterior (W) is obtained. Angiographic AVM nidus volume
length
after
correcting
intravoxel dephasing on size, may be evident
for magnification. effects show on multiple
by
are not incarotid
is calculated (c) Two-dimensional GRE
the matrix (arrow) sections, although
of the AVM only one is
to
seen here. (d, e) Contiguous two-dimensional GRE images obtained with flow compensation. Both sections include portions of AVM nidus. Flow-related enhancement accentuated by compensatory gradient moment nulling is isointense or hyperintense in comparable regions identifled on the non-flow-compensated MR image (illustrated in c ) as part of AVM nidus. Length (L) and width (W) are determined by means of direct measurement, Efferent pial veins
Volume
180
and
height
(v in e), although
#{149} Number
3
is approximated enhancing,
by multiplying are not included
section thickness in these volume
by number measurements.
of sections
containing
evidence
of AVM
Radiology
matrix.
#{149} 787
posttreatment able were Nine
MR image excluded from
of the
26 patients
was availthe study.
nidus
underwent
two follow-up MR imaging examinations, ii underwent three examinations, five underwent four examinations, and one underwent five examinations. The anticipated therapeutic effect of stereotactic radiosurgery is a reduction in residual active nidus volume over 12-24 months. This decline in nidus size, although progressive, is often nonlinear. In most cases, little involution occurs within the first 6 months, barring acute venous thrombosis, as can occur when venous aneurysms are present (4-7).
Roughly 40% of cases proceed to nidus thrombosis within the first year, and another 40% develop a thrombus within 2 years. Some diminution in size is likely in nearly every patient by 2 years,
even
when
complete
oblitera-
tion has not been achieved. The concept of active residual nidus volume is emphasized, as postradiosurgical healing results in a mixed gliotic and fibrotic response, which is evident on MR images as regions of low signal intensity within the original nidus territory. Residual active nidus volumes are therefore determined by signal-intensity
changes
between
se-
rial GRE studies, as described in Patients and Methods. For the purpose of this analysis, serial MR images were considered as either an appropriate or inappropriate reflection of the anticipated reduction in nidus volume after radiosurgery. An appropriate imaging result was assigned when no change or a progressive
diminution
in nidus
vol-
ume was observed on the follow-up images compared with earlier images. An inappropriate label was assigned if an increase in nidus volume was observed
on
a later
image
compared
with preceding images. In essence, serial MR images indicated either no change or a progressive diminution in nidus volume, depending primarily on the interval from their treatment date in all patients studied. None of the cases demonstrated an increase in nidus volume in a treated area or enlargement of the nidus in an untreated area.
(9,20).
DISCUSSION Surgical
excision
788
#{149} Radiology
of AVMs
requires
about the arterial contribuof brain along travel. Embolizaknowledge of
characteristics,
that
the nidus has arteriocapillary In other
words,
is,
a complex barrier
14
or
larly
after
radiosurgical
in size
as they
regress
ating a resultant ity artifact that time-of-flight
magnetic can simulate
As such,
it may
residual fast-flow nents often seen malformations.
across
nidus,
cre-
susceptibilfast-flow, the
BETWEEN
Figure
5.
MR
20%
15%
t.#{248}us SiZE ANGlO
25%
VARIANCE
AND MR
imaging-angiography
van-
ance. Nidus volume was overestimated in five patients and underestimated in seven with MR imaging; findings of MR imaging correlated with those of angiography (ANGlO) in seven patients.
100
50 40 30
E
20
underestimate
(fistulized) in residual Phase
shift
methods)
compovascular imaging
can
10
3
6
Figure
6.
12
9
MONThS
POST
Progressive
apparent
decline
at follow-up
patients
with
15
18
21
in nidus
MR imaging
angiographic
24
TREATMENT
size
in seven
evidence
of AVM
thrombosis.
nidus,
may (in its current state of development) be adequate for a pretherapeutic diagnosis of arteriovenous shunting in AVMs, but until improvements in spatial resolution and processing time develop, it is unlikely that small residual posttreatment nidus volumes can be accurately predicted. MR angiography (two-dimensional, three-dimensional, and dephase rephase
I 10%
ESTIMATED
8
particularly with asymmetric T2weighted and standard GRE sequences. For these reasons, several special measures to detect residual nidus flow have been devised. Phase shift imaging is not widely available but does provide a slowly obtained, relatively coarse image that lends itself best for slower-flow imaging (2123).
I
sequences,
in the
effects
0
(after
and MR angiography for that matter, may suggest persistent AVM because of the observation of persistent enlargement of the vascular pedicle. In addition, as thrombosis occurs, hemoaccumulates
2
75
of the afferent (Fig 8d) often routine Ti- or
spin-echo
siderin
24
treatment,
treatment), the ectasia and efferent circulation persists. Consequently, T2-weighted
12 10
requires special consideration. Although vascular maiformations diminish
a.
embolization
therapy requires knowledge of whether the nidus is truly a malformation (complex arteriocapillary barrier) or part of an arteriovenous fistula (simplified arteriocapillary barrier) and whether venous aneurysms are present. These structural aspects of vascular malformations can be defined with standard MR imaging techniques in multiple planes, often in conjunction with MR angiography (three-dimensional or dephase and/or rephase techniques). However, delineation of the extent of residual blood flow through an AVM nidus, particu-
and/or
precise information source(s) of afferent tors and the surfaces which these vessels tion therapy requires
flow
whether simplified
image
flow in both arterial and venous structures. These techniques for posttreatment AVM imaging have several disadvantages. Because the AVM residua reside along the outer aspects of the prior AVM nidus, it is useful to image both the thrombosed (ie, stationary) and nonthrombosed portions of the
nidus
for orientation. This relationis lost with all MR angiographic subtraction techniques. Both two- and three-dimensional techniques use short repetition times and depend on orthogonal blood flow relative to the section-select orientation to provide in-flow (time-of-flight) ship
signal intensity maximal-signal-intensity
for
subtraction and projection
methods of display. Smaller vessels, both arterial and venous (which form the AVM matrix), tend to reside nearly in-plane; they create less signal-intensity with adjacent
difference
compared
tissue, particularly at short repetition times, and therefore are frequently subtracted from the image. Both two- and three-dimensional MR angiographic methods tend to image the larger afferent and efferent components of vascular malformations rather than the nidus per Se. In addition, subacute blood products can sufficiently shorten Ti to artifactually appear as flowing blood at three-dimensional
MR
angiography.
Finally, in the late posttreatment period, residual ectasia of former afferent or efferent vessels can persist. Because
the
nidus
is often
not
seen
September
1991
compared with the section-select and phase directions). Gradient moment nulling techniques work best when flow-related enhancement is maximized. A velocity window is created on the basis of the relationship of section thickness to repetition time. The parameters in this study evolved from empiric observations made in an attempt to maximize velocity-dependent flow-related enhancement within the matrix of the vascular malformation. Difficulties immediately arise during attempts to quantify flow velocities within cerebral AVMs. These lesions
C.
typically
display
flow-rate
heterogeneity between patients and even within each AVM. A rough approximation of flow velocity is available from contrast agent clearance times
d. Figure 7. Flow-compensated enhancement of residual nidus in presence of subacute thrombus. (a) Axial GRE image obtained with flow compensation. Image was obtained 18 months after treatment but 2 months after development of spontaneous partial thrombosis in efferent AVM circulation (ie, clotted venous varix). Comparison of GRE images without (not illus-
trated) and with flow compensation eral aspect of matrix (arrow) and was
considered
residual
flow
and
revealed no intensity latter,
clear evidence change along
residual
thrombus.
of enhancement along left aspect (arrowhead). Flow-through
residual
right latFormer nidus
(ar-
row) was documented angiographically (see b). (b) Lateral vertebral angiogram obtained after MR image shown in a. Residual transnidus blood flow (arrow) is evident in the dorsal mesencephalic region, confirming what was predicted with MR imaging (see a). (c) Axial GRE image obtained with flow compensation 22 months after treatment. This section was obtained in the same manner and at the same level as a, but 4 months later. It demonstrates persistent hyperintensity compared with the residual clot on the left (arrowhead) but no marginal enhance-
ment on the right (arrow) lateral aspect. MR imaging findings were indicative of complete AVM thrombosis, which was confirmed angiographically (see d). (d) Lateral vertebral anglogram obtained after MR image shown in c. No residual transnidus blood flow can be detected, as correctly
predicted
by images
with
and
without
better with MR angiography, such techniques may suggest residual flow even after nidus obliteration has occurred. Such MR angiographic techniques were not available across the entire study period, creating an incomplete data set. Comparison of MR angiography with the AVM MR imaging protocol presented herein will be the subject of further investigation. Serial two-dimensional GRE imaging combined with flow compensation techniques in the manner described earlier has remained our method of assessment of persistence Volume
180
#{149} Number
3
flow-compensated
GRE
imaging
protocol.
of the arteriovenous shunt across the AVM nidus. This method has some strengths and some weaknesses, as do all the noninvasive means to assess residual intracerebral blood flow in AVMs. Gradient techniques, for
prove
signal
ity-dependent
moment the most
lost
nulling part, im-
to intravoxel,
phase
shifts
velocsecondary
to nonstationary nuclei. They are best suited to correct the phase shifts associated with in-plane linear (intravascular) flow in the readout direction (due to the longer period during which the readout gradient is in effect
observed
during
microcatheter-
izations of brain AVMs before embolization therapy and from transcranial Doppler examination. Reported distal middle cerebral artery flow rates range from 30 to 50 cm/sec in adults who have undergone endarterectomy (24). Because flow velocity is inversely proportional to the relationship between the squares of the diameters of the arteries and because it is not uncommon for the afferent and efferent AVM vessels to dilate threefold, the resulting flow velocity in both the afferent and efferent limbs (secondary to shunting)
across
the
AVM
might diminish ninefold resulting in effect flow cm/sec.
These
lated ties
observations
approximations prompted
GRE
imaging
study.
These
the
matrix
to iO-fold, rates of 3-5 and
of flow selection
parameters parameters
calcu-
velociof the
used
in this
provided
a
velocity profile that best captured the flow-related enhancement within the nonfistula components of most cerebral AVMs imaged in this study. Such transnidus flow rates are at best an approximation and are probably low. In a presentation at the i990 RSNA scientific assembly, Turski et al (25) reported data from phase-contrast MR imaging techniques in untreated cerebral AVMs that approximated afferent circulation flow rates at 200 cm/sec and venous flow rates at 20 cm/sec.
These
techniques,
in all likeli-
hood, measure different aspects of transnidus flow and help determine flow rates in patients in different clinical states. It is likely that radiosurgery may have a pronounced effect on overall transnidus flow rates. Nonlinear flow generally obviates the flow-related enhancing effects of gradient moment nulling. However, as observed
in this
study,
transnidus
Radiology
#{149} 789
flow is sufficiently in-plane and ciently linear to create detectable nal-intensity
differences
between
suffisigse-
rial GRE images that vary only in the application of the gradient moment nulling techniques. As one would expect, all flow-related phase shifts across the nidus are not corrected; neither are all corrected uniformly. However, the clinical object of such MR flow imaging is to detect the presence of any residual flow across the irradiated nidus and to detect whether there is residual flow outside the treatment field. The latter may occur if the margins of the original nidus were not exactly appreciated on the initial angiogram and were not encompassed by the radiation field. Data from this series indicate that signal-intensity change from hypointense (relative to cerebral parenchyma) to either isointense or hyperintense within the nidus is indicative of transnidus flow. The smallest region of residual transnidus flow was 1.5 mm
in cross-sectional
diameter.
(assuming
Ti-weighted
pulse
sequences, as described previously) performed during this intermediate phase of treatment will reveal a central region of low signal intensity (secondary to gliosis, fibrosis, and hemosiderin deposition) and evidence of marginal flow-related enhancement on GRE studies with and without flow compensation. These features are illustrated in Figures 7 and 8. The imaging of thrombosed portions of an AVM with GRE imaging is very helpful to identify residual flow once significant nidus reduction has occurred. These morphologic data are lost with MR angiography. Additional problems arise if intervening thrombus formation occurs within the matrix of the malformation. Acute thrombus most commonly forms within venous aneurysms associated with the efferent drainage system. The presence of blood products will generally add hyperintense signal intensity to the central and efferent portions of the nidus. It has been our experience in a few cases that flow-compensated GRE techniques, despite the presence of subacute 790
#{149} Radiology
C.
In
cases in which no residual flow was proved angiographically, no signalintensity change was observed in any part of the prior AVM matrix. After radiosurgical treatment, cerebral AVMs progressively decrease in size, becoming more gliotic and fibrotic. These changes generally appear centrally. As a consequence, the residual nidus tends to reside along the periphery of the AVM matrix. MR imaging
a.
b.
d.
Figure
8.
MR
imaging-angiographic
over 20 months. ment demonstrates rowheads) anterolaterally,
obtained
and
evidence which
anterolateral
treatment
gram
seen
radiosurgery of original
obtained blood
confirms nidus.
at follow-up after
which
after
was
flow.
tery (arrow), formerly treatment, even after
posttreatment
note
part of afferent nidus has been
subtle
smaller, lenged.
residual
patent
the technique Our current
nidus
chalsug-
gests that persistent residual transnidus flow can be reliably detected as
the 1.5-2
residual patent nidus mm in cross-sectional
(c) Axial
approaches diameter.
no
MR imaging residual
longer
ectasia
of at least
completed
carotid
anglo-
of residual
one
suprasylvian
if the
of the
AVM
flow
indicate
ar-
is not uncommon
Conceivably, residua
pre-
with enhance-
absence
of ectasia
(arrow)
from
displays
d). (d) Lateral
confirms
nidus
obtained
(a). MR features (see
nidus flow angiogram
reduced
GRE image
Section
image
AVM
size
after treatof AVM (ar-
residual carotid
substantially
pedicle. This degree thrombosed.
becomes
is further experience
of nidus
12 months portion
of residual
was
angiographically
blood products, allowed detection of persistent marginal blood flow, as illustrated in Figure 7. Results of the study indicate the accuracy of this method in determining transnidus flow. Certainly, flow persistence within an AVM matrix larger than 5 mm in cross-sectional diameter is relatively easy to detect. As the
size
treatment.
posttreatment
However,
the presence
MR imaging.
confirmed
diminution
(arrow), confirming in b. (b) Lateral cerebral
Nidus
radiosurgical
20-month
of progressive
with flow compensation representing thrombosed
enhancement proved
on the 12-month
thrombosis,
transnidus
after aspect
20 months
(arrow)
nidus
of peripheral is angiographically
size, as predicted
compensation
ment
(a) Axial GRE image obtained central low signal intensity
12 months
in the
confirmation
after
posttreatment undergo
fistuliza-
tion, the flow rate across the fistula could exceed the velocity window of the technique, and thus no signal would be captured with use of this method. This has not occurred in our series
the
to date.
nidus
source
of error,
ther
as more
dus
obliteration
ever,
our
fistulization
a likely it will
cases
be
with
studied
fur-
complete
ni-
How-
to date
indicates
flow-compensated with a reasonably
of reliability,
of
eventual
accumulate.
experience
that tandem, studies can,
degree
Because
remains
predict
GRE high
transni-
September
1991
dus flow or lack of flow in most all complex matrix AVMs. U
apy: the role of MR. Neuroradiology
if not 9.
MR imaging
References 1.
2.
tentonal
Marks MP, Delapaz RL, Fabrikant JI, et al. Intracranial vascular malformations: imaging of charged particle radiosurgery. I. ResuIts of therapy. Radiology 1988; 168:447455. Marks MP, Delapaz RL, FabrikantJl, et al. Intracranial vascular malformations: imaging of charged particle radiosurgery. II. Complications. Radiology 1988; 168:457-
10.
11.
462.
3.
4.
Fabrikant JI, Lyman JT, Frankel KA. Heavy charged particle Bragg peak radiosurgery for intracranial vascular disorders. Radiat Res 1985; 8(suppl):S244-S258. Lunsford LD, FlickingerJ, Coffey RJ. Ste-
reotactic gamma North American 5. 6.
7.
Arch Neurol 1990; 47:169-175. Friedman W. LINAC Radiosurgery. Neurosurg Clin North Am 1990; 1:991-1008. Wolkov HB, Bagshaw M. Conventional radiation therapy in the management of arteriovenous malformations of the central nervous system. Int J Radiat Oncol Biol Phys 1988; 15:1461-1464. Levy R, FabricantJ, Frankel K, Phillips M, Lyman J. Stereotactic heavy charged partide bragg peak radiosurgery for the treat-
ment
8.
knife radiosurgery: initial experience in 207 patients.
of intracranial
arteriovenous
mations in childhood and adolescence. Neurosurgery 1989; 24:841-850. Jabour BA, DionjE, Lufkin R, et al. rovascular lesions and endovascular
12.
#{149} Number
3
H, Van
Intracranial and clinical
Y, et al.
R, Wentz
K, Mattle
H, et aL
In-
arteriovenous
malformations:
evaluation
with
MR angiography 1989; 173:831-
Radiology
a special course in MR 1990. Oak Brook, Radiological Society of North America, 1990; 57-62.
Ill:
Atlas
RI.
AS, Fram
Neuther-
Needell
WM, Maravilla
aging
in vascular
dient
recalled
9:637-642. Waluch V.
KR.
MR flow im-
malformations
acquisition. Magnetic
flow. Semin
AJNR resonance
Neurol
using 1988;
Dawson Treatment
malformations:
A,
D. MR
1985; 156:383-
T, Yamada
R III, Tarr R, Hecht of arteriovenous
11:857-864. Dumoulin C, Souza Three-dimensional
raphy. 22.
23.
25.
EK, Grossman
L, Uske
G, Norman
Radiology
of the brain with combined and stereotactic radiosurgery.
Masaryk T. MR Angiography. In: Kressel HY, Modic MT, Murphy WA, eds. Syllabus:
1988; 169:455-461.
16.
20.
24.
SW, Mark
vascular
S. Nishimura
Laub
Magn
Reson
G, Kaiser
W.
gradient
tracerebral
selective
389. Imakita
Pleghos
M, Dooms
N, et al.
Cerebral vascular malformations: applications of magnetic resonance imaging to differential diagnosis. Neuroradiology 1989; 31:320-325.
MR angiogra-
Vascular intracranial lesions: applications of gradient echo MR imaging. Radiology
malfor-
19.
Fraeyenhoven
837.
15.
W, Lemme
and CT imaging.
1990; 175:443-448.
and venography. 14.
Kucharczyk Intracranial
of supraAmJ Roent-
vascular lesions: optievaluation of three-
time-of-flight
phy. Radiology Edelman
18.
21.
211-220. Marchal G, Bosmans
of blood
180
Kikuchi
genol Souhami L, Olivier A, Podgorsak EB, Hazel I, Pla M, Tampieri D. Dynamic stereotactic radiosurgery in arteriovenous malformation: preliminary treatment results. Cancer 1990; 66:15-20. Edelman RR, Mattle HP, Atkinson DJ, et al. Cerebral blood flow: assessment with dynamic contrast enhanced T2*weighted MR imaging at 1.5 T. Radiology 1990; 176:
mization dimensional
17.
Volume
CM,
in the management
intracranial AVMs. 1988; 150:1143-1153.
L, et al.
13.
1989;
Brant Zawadzki
31:341-345. Smith HJ, Strother
S. et al. malformations
embolization AJNR 1990;
S, Walker M, Wagle W. phase contrast angiog-
Med 1989; 9:139-149. MR angiography with refocusing. J Comput As-
motion
sist Tomogr 1988; 12:377-382. Lenz G, Haacke E, Masark T, Laub
plane vascular imaging: pulse design and strategy. Radiology
G. Insequence 1988; 166:
875-882. HalseyJ, Morawetz
H, Gelmon
S.
velocity
in the
McDowell R.
Blood
middle
cerebral artery and regional cerebral blood flow during carotid endarterectomy. Stroke 1989; 20:5358. Turski PA, Korosec FR, Partington CR, Strother CM, Graves VB, Mistretta CA. Cardiac-gated and variable-velocity-phase MR angiography for evaluation of intracranial aneurysms and arteriovenous malformations (abstr). Radiology 1990; 177(P):281.
gra-
imaging
1986; 6:65-71.
Radiology
#{149} 791