Raad David
H. Mohiaddin, N. Firmin,
Vena with
#{149} Samuel L Wann, #{149} Simon Rees, FRCR
MD PhD
with
right-sided
cardiac
disease. In the control subjects, peaks of flow in systole and diastole were observed, and mean SVC flow was 35% of the cardiac output. Respiratory gating was used in six control subjects to acquire images at end inspiration and end expiration, and although the systolic peak was reduced at end expiration, total flow was unchanged. A reduced systolic peak and retrograde flow in the IVC were observed in patients with tricuspid regurgitation. A reduced diastolic peak was seen in patients with pulmonary
hypertension,
pericardi-
al constriction, and right ventricular dysplasia, reflecting reduced diastolic compliance of the right yentricle. In the patient with obstruction of the SVC, absence of flow was confirmed, and retrograde flow was seen in the azygos vein. The authors believe that cine MR velocity mapping is a reliable method of studying vena caval flow noninvasively and that it has important p0tential applications for the investigation of disorders of the right side of the heart. Index
terms: Blood, flow dynamics, 524.1299 Blood, MR studies, 51.1214, 566.1214, 569.1214 Heart diseases, 51.781, 53i.84 #{149} Heart, flow dynamics, 524.1299 #{149} Heart, MR studies, 51.1214 Magnetic resonance (MR), cine study #{149} Venae cavae, 566.91, 569.91 #{149} Venae cavae, abnormalities, 566.781, 566.83 #{149} Venae cavae, MR studies, 566.1214, 569.1214 Radiology
#{149} Richard
Underwood, B. Longmore, FRCS
1990;
#{149}
#{149}
V
ENOUS
blood
flow
to the
heart
has been a subject of interest for centuries, dating from Erasistratus (circa 300 B.C. to circa 250), Galen (circa AD. i29 to circa A.D. 200), and Wilham Harvey (i578-i657) (i). Because the heart can pump only the blood that it receives, the volume of blood returning to the heart is an important determinant of cardiac output. The mechanisms facilitating venous return are therefore of considerable interest, but the roles of myocardial contraction, active cardiac suction, respiration, skeletal muscle contraction, and hydrostatic pressure in augmenting venous return and ventricular filling remain controversial. The pattern of central venous blood flow is also important in the clinical evaluation of conditions such as constrictive pericarditis, cardiac tamponade, restrictive myocardial processes, tricuspid regurgitation and stenosis, pulmonary regurgitation and stenosis, and congestive heart failure. Central venous blood flow can be studied by means of several invasive and noninvasive techniques. At thoracotomy, flow can be measured by placing flow meters around or into the caval veins (2-4). Alternatively, specialized catheters can be used to measure both pressure and flow percutaneously (5). Bedside observation of the jugular venous pulse is the simplest noninvasive technique, and recordings of the jugular venous pulse contour approximate to the right atrial pressure curve (6). Doppler echocardiography of tricuspid inflow and hepatic venous flow have been used to record abnormal blood flow velocities in patients with re-
177:537-541
I From the Magnetic Resonance Unit, National Heart and Chest Hospitals, 30 Britten St. London SW3 6NN, England (R.H.M., RU., D.N.F., SR., D.B.L.); and the Medical College of Wisconsin, Milwaukee (S.L.W.). Received January 24, 1990; revision requested March 14; revision received June 14; accepted June 15. Supported in part by the Board of Governors of the National Heart and Chest Hospitals, the Coronary Artery Disease Association, and Picker International Ltd. Address reprint
requests
e RSNA,
to R.H.M.
1990
#{149}
MRCP
Caval Flow: Assessment Cine MR Velocity Mapping’
The authors used cine magnetic resonance (MR) velocity mapping to study flow in the superior vena cava (SVC) and inferior vena cava (IVC) of 13 healthy control subjects and 13 patients
FACC
#{149} Donald
strictive processes (7). Although useful, none of these methods is ideally suited to the volumetric analysis of central venous blood flow. In this artide we describe a new noninvasive method of measuring blood flow in the inferior vena cava (IVC) and superior vena cava (SVC) with cine magnetic resonance (MR) velocity mapping.
SUBJECTS
AND
METHODS
Subjects Thirteen
healthy
control
subjects
and
13 patients with heart disease known to affect the pattern of central venous filling were studied. The healthy control subjects were all men and had no history, symptoms, or physical evidence of heart disease. Their mean age was 36 years (mange, 31-50 years). The patients are listed in Table 1. Tricuspid regurgitation was documented at both echocardiography and cine MR imaging of the tricuspid valve. Pulmonary hypertension was confirmed by measurement of pulmonary arterial pressure during catheterization. Morphologic changes in the right ventricle secondary to pulmonary hypertension were demonstrated with MR imaging. The diagnosis of arrhythmogenic right ventricular dysplasia was based on electrophysiologic studies coupled with evidence of right ventricular regional abnormalities
MR images (8). Constrictive was confirmed by the pres-
on
pericarditis
ence of clinical symptoms and signs coupled with evidence of pericardial thickening at echocardiography and MR imaging. All patients were in sinus rhythm.
MR An
Imaging MR
International,
imager
(Vista
MR
Highland
2055;
Heights,
Picker Ohio)
operating
at 0.5 T was used with a surface receiver coil. A spin-echo (SE) sequence (echo time [TE], 40 msec) was used with
Abbreviations: = spin echo, echo time.
IVC SVC
=
=
inferior
superior
vena vena
cava,
cava, TE
SE =
537
electrocardiographic multiple transverse
the
heart.
and
the
gating images
Section field
thickness
of view
resolution
of 256
encoding
direction
was
was in
and
the
step
was
with
a
Age
in the
Each
acquired
phase-en-
twice
and
aver-
aged so that imaging time was between 3 and 4 minutes for each set of images, depending cine MR
on heart velocity
rate. The mapping
from these images at the furcation of the pulmonary
SVC and IVC with Cine
for identical
of the biartery for
2 cm below the junction the right atrium. MR
formed sequence
planes were level
velocity
mapping
with a field even (TE = 12 msec)
of velocity
plane nance
in the signal.
phase The
ously
described
was with
encoding
flow measurements curate to within
resoprevi-
validated,
and left healthy
the
previous
ventricular subjects.
maps were velocity-encoded
erence and that
paragraph.
without
velocity
acquisition time of a conventional
in 6-8
heart
nate.
regularities difficult and images be obtained.
of atnial However,
were extended cardiac cycle 95% of the in the SVC
the mean cycle and
ir-
gating cycle, not
over as large a pant of the as possible, and, typically, was imaged. calculated
Flow from
the cardiac area of the
vein. Cardiac
output
was
measured
of cine gradient echo imaging zontal and vertical long-axis
by
means
in the planes
honi-
calculated from the left ventricular stroke volume and the heart rate. The severity of tricuspid regurgitation was judged from the size of the regurgijet
signal
seen loss
on did
cine not
MR
images
extend
into
(14).
Because
the
venae
Performed
and Gating
constraints
Cardiac gating alone was flow in the SVC in seven
hypertension
(59/26
Used
in 13 Control
Study
mm
mm
Hg)
regurgitation regurgitation
Subjects
and
13
Gating
No.
SVC
IVC
Cardiac Output
7 6 10
Yes No Yes
No Yes No
Yes No No
Yes Yes Yes
No Yes No
3
Yes
Yes
No
Yes
No
were
made
in the SVC and
subjects and in the IVC in the the former group, left ventricular curves were also tical and horizontal
generated long-axis
of the left ventricle. both respiratory and
In the cardiac
used to compare vena caval inspiration, at end expiration, out was
respiratory gating. used to generate
and part
end inspiration of the cycle
chest fined
expansion. similarly
were
from the yencine images latter group, gating were at end with-
A pneumograph a respiratory
flow
trace,
defined as that 30% of maximum
within
and
recorded
expiration the minimum.
the
six. volume
and
was End about
other
was
deThe
electrocardiogram
simultaneously
to docu-
ment the accuracy of gating. Each flow study with respiratory gating required approximately 30 minutes. In the 13 patients, cardiac
vena caval flow gating alone.
Statistical
imaged
with
Analysis
Statistical an
was
analysis
analysis
was
of variance,
Student
t tests
performed followed
with by
as appropriate.
RESULTS of imaging
time, respiratory gating was not used in all subjects. Each patient and control subject were studied only once, but the study protocol differed. Table 2 shows the site of flow study and the type of gating used.
#{149} Radiology
Studies
Note-Flow measurements of the left ventricle.
Techniques of the
pulmonary
IVC. Cardiac
Cardiac
output
Respiratory
was measured
from
cine
images
paired
Gating
2 of Flow
Patients
The
cavae.
538
Right ventricular dysplasia Right ventricular dysplasia Acquired immunodeficiency syndrome, pericardial constriction Cardiac sarcoidosis Pulmonary hypertension (unknown cause), moderate tricuspid Ischemic heart disease, congestive heart failure, severe tricuspid Severe tricuspid regurgitation (cause unknown) Moderate tricuspid regurgitation, cardiomyopathy Rhabdomyosarcoma, SVC obstruction
pneumogram
through the left ventricle with the imaging parameters described above. Diastolic and systolic left ventricular volumes were measured with a biplane area-length technique (13), and cardiac output was
tant
19/M 28/M 36/M 36/M 32/F 59/M 62/M 64/M 62/M
thromboembolism,
on other
cardiac cardiac
velocity throughout the cross-sectional
Pulmonary
Control
a a ref-
contraction could the acquisitions
cardiac cycle and IVC was
28/F
twice image ob-
and
made reliable at the end of the
Pulmonary sarcoidosis, pulmonary hypertension (55/25 mm Hg) Usual interstitial pneumonia, pulmonary hypertension (60/26 mm Hg) Pulmonary lymphangioleiomyomatosis, pulmonary hypertension (59/27 Hg)
velocity
depending
arrhythmia
Disease
62/M 36/F 45/F
Subjects
encoding,
was therefore anatomic
minutes,
Sinus
Heart
Diagnosis
Flow
Tem-
by subtracting image from
phase
image
tamed
The
constructed
with
Patients
poral resolution was 1 6 frames per cardiac cycle, and the imaging parameters were the same as for the SE sequence described in
13 Patients
(y)/Sex
Table Types
and
were shown to be ac5% (1 1,12) by a compani-
flow in
of the
imaging
of the magnetic technique was and
Details
per-
rephasing
to the
(9,10)
son of aortic stroke volume
of the
echo
perpendicular
1
Clinical
frequency-
128 pixels
direction.
Table
10 mm
30 cm,
pixels
phase-encoding coding
to acquire encompassing
used to image of the control
Control
Subject
Studies
Figure 1 shows representative images and flow measurements in a control subject with cardiac gating alone. The flow pattern in the SVC was qualitatively similar to that in
In
the IVC, but the volume of flow was less (32% of total flow in this example). Systolic and diastolic peaks were always seen, the former being larger than the latter in both venae cavae. The IVC flow patterns observed with respiratory and cardiac gating in the six control subjects were similam to those observed with cardiac gating alone, although flow at end inspiration was higher than at end expiration, and flow measured with cardiac gating alone closely resembled flow at end expiration (Fig 2). Mean peak systolic IVC flow was 9.6 L/min :E 1.8 (standard deviation) with end-expiratory gating, 6.6 LI mm ± 1 .8 with end-inspiratory gating, and 8.4 L/min ± 2.4 with cardiac gating alone. Mean peak diastolic Ivc flow was not significantly different at end expiration (7.8 L/min ± 1.8), at end inspiration (5.4 L/min ± 1.8), or with cardiac gating alone (6.6 L/min ± 1.2). There were no statistically significant differences among the values recorded with the three gating techniques. Mean IVC flow for the entire cardiac cycle was also not significantly different with endexpimatory
gating,
end-inspimatory
gating, or cardiac gating alone: 3.1 LI mm ± 0.7, 3.1 L/min ± 0.5, and 3.0 L/min ± 0.5, respectively. The cross-
November
1990
..
.,.t,-:F
&
.
I.’ 4-,
.
,
,4,,
‘
.4,
,
.
‘1 ..
-.4
.,
,
,-., ‘‘r. V
r..:
-
,,
..,,
S
1.
;
.:
k
,
b.
a. Figure
10
8
-J
ities
4
0
(a) SE image
of the SVC (arrows)
artery and three frames from the and middiastole (3). The velocity ties in lighter shades of gray, and and middiastolic peaks of flow in in the ascending aorta in 1 because
C
E
1.
expected
in
the
venous
system.
at the level
of the bifurcation
of the pulmonary
cine velocity map display in midsystole (1), end systole (2), maps indicate zero velocity as medium gray, caudal velocicranial velocities in darker shades of gray. The midsystolic the SVC are seen. There is aliasing of the velocity display the sensitivity of the acquisition was set to the low velocNote
also
the
complex
spiraling
of
flow
in
the
right
pul-
the backflow channel in the ascending aorta in 2. AA ascending aorta, DA = descending aorta, RPA right pulmonary artery. (b) Images corresponding to a show the IVC (arrows) 2 cm below its junction with the right atrium. The systolic and diastolic peaks of flow are seen. (c) Vena caval flow throughout the cardiac cycle calculated from the complete cine acquisition at each site. monary
11. 2
0
Time
artery
and
(ms)
C.
SVC ular
12.0 9.0
-
6.0
-
End
Exp
End
Insp
...
flow plotted against left ventnicvolume is shown in Figure 3.
CG alone
C
E
Patient
Studies
:1 3.0 0 U-
0.0 .30
0
200
400
Time
600
800
(ms)
Figure
2. Flow in the IVC gated to end expiration (End Exp) and end inspiration (End Insp), and with cardiac gating (CG) alone. Peak flow in systole is greater at end expiration, but total flow throughout the cycle is the same with all three methods of gating.
sectional area of the IVC varied considerably from end expiration (6.9 cm2 ± 0.3) to end inspiration (4.9 cm2 ±
0.5).
Mean SVC flow (1.8 L/min ± 0.3) was 35% ± 3 of cardiac output measured from the left ventricular stroke volume in six subjects. An example of
Volume
177 #{149} Number
2
In the patients with tricuspid megurgitation the systolic peak of flow was markedly reduced to the extent that retrograde flow occurred in some cases (Fig 4). Pulmonary hypertension, penicardial constriction, and myocardial restriction had the opposite effect, with attenuation of the diastolic peak (Fig 5). In the patient with a large rhabdomyosarcoma, obstruction of the SVC was confirmed from the absence of flow within it and from the retrograde flow in a dilated azygos vein (Fig 6).
DISCUSSION We have demonstrated the ability of cine MR velocity mapping to measure blood flow in the venae cavae of healthy control subjects and of patients with abnormalities known to affect the pattern of central venous return. Normal patterns show systol-
ic and diastolic peaks, but in the presence of tricuspid regurgitation, maximal forward flow occurs during diastole, and when diastolic filling of the right ventricle is impaired, maximum flow occurs during systole (Fig 7). We have previously validated the techniques used in this study and have demonstrated that MR flow measurements in the aorta and pulmonary artery agree well with each other and with left and might ventnicular stroke volumes in healthy subjects (15-17). Others have shown a good correlation between MR flow measurement of venous (SVC and IVC) return and aortic flow and between aortic flow measured with this technique and Doppler ultrasound (US) (18). Although it would be desirable to measure flow in the venae cavae with cine MR velocity mapping and an alternative method, the constraints of currently available imaging systems make this comparison impractical. The main advantages of using cine MR velocity mapping for the measurement of vena caval flow are the
Radiology
#{149} 539
15
12
C
8
E
E
N
E
_J
C S
>
0
8
23 U.
Time (ms)
Time(ms)
Figure
3. Left ventricular (LV) volume measured from cine MR images by use of an area-length technique together with SVC flow in a control subject. The relationship of the peaks in flow to the cardiac cycle can be seen.
flme(ms)
5.
4.
Figures
4, 5.
(4) Flow
in the SVC and
IVC in a patient
tolic peak is attenuated, and retrograde flow occurs IVC flow in a patient with pulmonary hypertension. implies impaired right ventricular filling.
with
tricuspid
regurgitation.
The sys-
in the IVC during systole. (5) SVC and The diastolic peak is attenuated, which
12
p
10
T, 8
C
E
...4.
6 -j
4
4
0’
.
.?.-‘,‘4..:
0 U.
,
::1
.
.t,.
2 .
.
-,_,5
0
.
-2
:z 0
100
200
Time Figure
6. (a) SE image (top left) and velocity maps a patient with a rhabdomyosarcoma (fl and obstruction flow in a dilated azygos vein (solid straight arrows) flow with a normal pattern in the IVC, absent flow
noninvasive and the anatomic
nature of the technique fact that it provides excellent images as well as measureof
flow.
Invasive
techniques
are obviously less desirable, and the use of devices such as electromagnetic flow meters that encircle the veins may distort the shape of these veins and alter the flow. Catheter-tip flow meters can measure only the velocity and not the volume of flow and are subject to inaccuracies when flow patterns in the vessel are complex (Fig 1). Doppler echocardiogmaphy has the same disadvantage, and, although it is possible to measure the cross-sectional area of the venae cavae (19) and hence to measure volume flow, such measurements have not been reported to our knowledge, possibly because of the difficulties of 540
400
(ms)
b.
a.
ments
300
. Radiology
in early of
systole the
and normal in the SVC,
SVC.
(top
right),
Antegrade
flow in the and retrograde
midsystole flow
descending flow
in
(bottom the
left), (curved
and
of Respiration
We have shown some difference between vena caval flow at end inspimation and end expiration in control subjects, but less than we expect-
early
arrows)
aorta (open arrows). in the azygos vein. Desc
recording adequate cross-sectional images together with Doppler US measurements. The limitations of cine MR velocity mapping include the relatively long acquisition times, the confined bore of the magnet (which makes imaging of sick patients difficult), and the high cost of the imager. These limitations may become less important with the development of real-time MR velocity mapping that uses an echo planar technique (20), open-access magnets, and cheaper imagers.
Effects
IVC
diastole is seen,
(b) Ao
(bottom as are
Flow curves descending
=
right)
in
retrograde
showing aorta.
high
ed. Several factors may be responsible. First, the images were acquired at end expiration and end inspiration, and the changes in intrathoracic pressure at the extremes of the respiratory cycle may not be as great as those in the middle of inspiration or expiration. Second, the gating system available to us allowed sampling during both phases of respiration, and so both phases of the respiratory cycle were included at each extreme. Third, it is possible that respiration does affect the velocity of blood in the caval veins but that the cross-sectional area also changes so that the total volume of flow remains relatively constant (21). We certainly observed a considerable difference in diameter at end expiration and end inspiration. Fourth, the effects of res-
November
1990
onance.
Healthy people are known to spend more time exhaling than inhaling (22). This may explain why flow measured without respiratory gating
c
, FLOW
‘.7
Tricusptd
regurgttatJon
(,i
closely
resembled
flow
11.
at end
Chest
SR, Firmin DN, Klipstein RH, Longmore DB. Magnetic resonance velocity mapping: clinical application of a new technique. Br Heart J 1987; RSO,
Firmin
DN, Nayler
GL, Klipstein
RH, UnIn
derwood SR, Rees RSO, Longmore DB. vivo validation of MR velocity imaging. Comput Assist Tomogr 1987; 11:751-756.
13.
14.
Underwood SR, Gill CRW, Firmin DN, et al Left ventricular volume measured rapidly by oblique magnetic resonance imaging. Br Heart J 1988; 60:188-195. Sechtem U, Pflugfelder PW, Cassidy MM, et al. Mitral or aortic regurgitation: quantification
of regurgitant
cine MR imaging.
We thank the staff of the of the National Heart and
unit
Hospitals
for
their
15.
volumes
Radiology
with
1988;
Restricted
right
ventricular
1.
filling
c1
3.
assistance.
16.
Brecher
GA.
Venous
return.
New
York:
1956; 1-10.
& Stratton,
4.
Brecher GA. Cardiac variations in venous return studied with a new bristle flowmeter. Am J Physiol 1954; 176:423-430. Guntheroth WG, Morgan BC, Mullins GL. Effect of heart beat and respiration on
The effects of the mechanics side of the heart on vena caval
of
T.
Normal
superior
vena
Cotomy.
Scand
flow cava in man
J Thorac
Wexler
L, Bergel
DH,
Mills CJ. Velocity mal human venae
pattern during
17.
18.
in the thora-
Cardiovasc Gabe
Sung
IT, Makin
177 #{149} Number
2
6.
7.
8.
Tavel
ME.
Clinical
Underwood
of blood flow in norcavae. Circ Res 1968; phonocardiography
Rowlands
E,
Mohiaddin RH, Rees RSO, Longmore Global and regional right ventricular
DB.
function
SR,
in
Burman
ED,
arrhythmogenic
right
ventric-
ular dysplasia studied by magnetic resonance imaging (abstr). Magn Reson Imaging 1990; 8(suppl l):248. 9.
10.
Bryant
DJ, Payne
JA, Firmin
DN,
RH,
Firmin
Long-
more DB. Measurement of flow with NMR imaging using a gradient pulse and phase difference technique. J Comput Assist Tomogr 1984; 8:588-593. Nayler GL, Firmin DN, Longmore DB. Blood flow imaging by cine magnetic res-
DN,
et
and metroaorta
velocity
by
mapping.
117:1214-1222.
HG, Klipstein Pulmonary
RH, Mohiaddin
RH,
distensibility
and
artery
flow patterns: a magnetic resonance of normal subjects and patients with pulmonary arterial hypertension. Am Heart J 1989; 118:990-999. Firmin DN, Nayler GL, Klipstein RH, Underwood SR, Rees RSO, Longmore DB. In validation
of MR velocity
imaging.
Comput Assist Tomogr 1987; 11:751-756. Van Rossum A, Sprenger KH, Peels FC, Visser FC, Valk J, Roos JP. In vivo validation of quantitative flow imaging in arteries and veins using magnetic resonance phase shift techniques (abstr). In: Book of abstracts: Society of Magnetic Resonance in Medicine 1989. Vol 1. Berkeley, Calif: Society of Magnetic Resonance in Medicine, 1989; 205.
GS,
and external pulse recording. Chicago: Year Book Medical, 1978; 250-266. Appleton CP, Hatle LK, Popp RL. Demonstration of restrictive ventricular physiology by Doppler echocardiography. J Am Coll Cardiol 1988; 11:757-768.
al.
vivo
23:349-359.
piration on vena caval flow are likely to be most marked in patients with restrictive or constrictive cardiac disease, and, for practical reasons, we were unable to use respiratory gating in the patients studied. Our observation that total vena caval flow was the same at end expiration and at end inspiration suggests a dominant effect of myocardial events in determining venous return. As believed by Erasistratus and Galen, but not by Harvey, the heart may act in part as a pressure-suction pump independently of the effects of respiration.
Klipstein
blood study
1972; 6:22-32. 5.
flow.
Froysaker
Bogren
et
flow patterns in the cavae, pulmonary artery, pulmonary veins and aorta in intact dogs. Science 1965; 150:373.
FLOW
HG,
magnetic resonance Am Heart J 1989;
Grune
2.
Bogren
al. Quantitation of antegrade grade blood flow in the human
References
Volume
1986;
57:404-412.
12.
:>y/NN\ P1.0W
the right
Tomogr
167:425-430.
MR imaging
7.
Assist
Underwood Rees
expiration, although a further explanation may be that the gating spanned both phases of the respiratory cycle, as explained. In any event, it appears that respiratory gating is not required for cine MR imaging measurements of vena caval flow. This is important because of the longer time required for respiratory gated acquisition. U Acknowledgment:
Figure
J Comput
10:715-722.
19.
Himelman Schiller
with sitive
RB, Kircher NB.
blunted
Inferior
respiratory
echocardiographic
tamponade.
B, Rockey vena
cava
DC, plethora
response: sign
J Am Coil Cardiol
a sen-
of cardiac
1988;
12:1470-1477.
20.
21.
22.
Firmin DN, Klipstein RH, Hounsfield GL, Paley MP, Longmore DB. Echo-planar flow velocity mapping in high resolution. Magn Reson Med 1989; 12:316-327. Permut S, Riley RL. Hemodynamics of collapsible vessels with tone: the vascular waterfall. J AppI Physiol 1963; 18:924-932. Shea SA, WalterJ, Murphy K, Guz A. Evidence for the individuality of breathing pattern in resting healthy man. Resp Physiol 1987; 68:331-344.
Radiology
#{149} 541