J.

Paul Finn, MD #{149} Robert R. Edelman, MD #{149} Roger H. Esterbrook Longmaid, MD #{149} Robert A. Kane, MD Heinrich P. Mattle, MD #{149} Melvin E. Clouse, MD

Liver with

Transplantation: Surgical Validation’

O

Thirty patients (mean age, 45 years) were evaluated with magnetic resonance (MR) angiography before liver transplantation to assess the accuracy of MR angiography. A series of breath-hold, two-dimensional images were acquired and subsequently processed to form three-dimensional projection angiograms. Graphic information on blood flow in the portal vein was acquired by using presaturation bolus tracking. Correlative duplex ultrasound (US) was performed in 28 patients, and surgical or autopsy correlation was available in all cases. MR angiography demonstrated patency of the portal vein in 26 (96%) of 27 patients, made possible the diagnosis of portal venous occlusion in three of three patients, depicted reversed portal flow in one patient, and provided clear delineation of the extent of varices and specific portosystemic collateral vessels. When duplex US was successful, there was full agreement with MR angiographic results in assessing portal vein patency and flow direction. All of the MR findings were corroborated at surgical exploration or autopsy. The authors conclude that MR angiography is very accurate in the portal system and is valuable in preoperative assessment for liver transplantation.

Index terms: Liver, transplantation. Portal vein, thrombosis, 957.751 MR studies, 957.1214 #{149} Magnetic (MR), angiography, 957.1214 Radiology

I

From

1991;

the

Departments

761.459. vein, resonance

#{149} Portal

of Radiology

(J.P.F.,

MR

MD #{149} W. David R. Stokes, MD

H.E.L.,

Deaconess Hospital, Israel Hospital, Boston 13; revision received

RTHOTOPIC

R.A.K.,

K.R.S.,

H.P.M.,

M.E.C.)

Lewis,

MD

#{149}

#{149}

Angiography

liver transplantation has gained wide acceptance as the treatment of choice for a variety of end-stage liver diseases. Stringent selection criteria initially applied to donor and recipient are being steadiby relaxed, and the number of transplantations performed annually in the United States is now above 1,600 (1). Most recipients have advanced parenchymal or cholestatic liver disease, and theme is a high prevalence of portal hypertension. When a patient is considered a candidate for eithem portosystemic shunt placement or liver transplantation, knowledge of host portal anatomy and hemodynamic characteristics is of vital importance to the surgeon in planning the operative approach. For these purposes, some institutions still perform panhepatic angiography (2,3), but the procedure is invasive and may not be wholly effective in demonstrating varices and specific portosystemic collateral vessels (4). Concurrent renal impairment can increase the risk associated with conventional angiogmaphy in this group of patients, and the procedure is not routinely performed in pretransplantation workup at our institution. It has recently been shown that time-of-flight magnetic resonance (MR) angiogmaphy gives excellent delineation of portal anatomy and blood flow (5). To our knowledge, however, the accuracy of these MR angiographic techniques has not been validated in a clinical setting. The aggressive process of liver transplantation, which necessitates extensive mobilization of the liver and

179:265-269

(R.L.J., W.D.L.), New England partment of Radiology, Beth revision requested November reprint requests to J.P.F. ©RSNA, 1991

L Jenkins, #{149} Kenneth

and

portal structures (6), presents a unique opportunity to confirm or mefute imaging findings. To establish the accuracy of MR angiography, we undertook to obtain surgical correbation with preoperative MR angiographic findings in patients undergoing liver transplantation surgery.

SUBJECTS

AND

METHODS

Subjects Thirty consecutive adult patients were studied with MR angiography and presaturation bolus tracking (5) during preoperative workup for liver transplantation. There were 21 men and nine women with a mean age of 45 years (range, 25-65 years). Twenty-seven patients had histologically

proved

cirrhosis:

12 due

to

chronic active hepatitis, eight due to alcohob abuse, four due to scberosing cholangitis, two due to hemochromatosis, and one due to primary biliary cirrhosis (Table). Two other patients had hepatoma, and one had cholangiocarcinoma. Patients were examined sequentially as they presented for surgical workup; none was excluded. Surgical correlation with the imaging findings was available in 29 cases, and autopsy correlation was available in one case. Imaging

Methods

The MR angiographic studies were performed on a 1.0-T whole-body imaging system (Siemens Medical Systems, Iselin, NJ) with a maximum gradient strength of 6 mT/rn. A time-of-flight MR angiographic technique was employed that involved acquisition of sequential, singlesection, flow-compensated, fast low-angle shot images (7), each during a breathhold interval. Sequence parameters were as follows: repetition time (TR) msec/ echo time (TE) msec, 30/10; flip angle, 300; one signal averaged; and a 256 X 256 or 256 X 192 matrix. Section thickness

Surgery

185 Pilgrim Rd. Boston, MA 02215, and the De(R.R.E., H.P.M.). Received September 14, 1990; December 21; accepted December 26. Address

Abbreviations: = echo time,

TR

MPV = main portal repetition time.

vein,

TE

265

was

5-10

mm,

overlapped

and by

acquisitions

were

sagittal indicated. in

the

for

selective

and

to better

sequence 40/ 10; flip

pulse selectable

generate ously

The less

of the

with slab

to the

vessel

magnitudes quency-encoding saturation.

a maxito

portal

by

acquired of 90

Figure shows

vein a

and 60 trackduring

on a 128 msec.

oriented

The

adjusting

the

relative

and freduring precoronal

displayed

to show In 28 was

graphically

in a cine

direction duplex

and pattern. ultrasound (US)

the flow patients,

performed

within

24

hours

Calif) color

made

probe Doppler

MR

for

US

quantitative

be made

from

flow

bolus

imwere

unfavorable All MR

portal velocity not routinely insonation angiographic

angles. images

ual sections and projection assessed preoperatively

due

of

teries

Radiobogic

266

and

.

veins.

Radiology

of intraabdominal

noted.

When

the

vein

and

varices

patent,

the

and hepatic reanastomosed

MPV,

the

were inferior

artery were tranduring trans-

Enlarged coronary,

liver

surgery,

portal

splenic, gonadal,

superior renal,

were tissue

gas-

noted planes,

signal

specimens

were

submit-

No size

would

whether

sive,

then

was vessels that

be prone

errors.

and

attempt

of these we believed If the

were

believed

they could of vanices

moderate

MR angiography in

to

to substantial surgeons

MR angiography predict the location

they

made quantitasuch a

or mas-

was judged

to

this regard. Any discrepansurgical findings and MR findings were noted.

(individ-

images) were without knowl-

interalso

of the hepatic veins the major visceral reports

aron

tab vein thrombus shown patients

occlusion. Nonocclusive in the portal vein was with MR angiography in two (Fig 2), and one patient had

a patent

MPV

tal

occlusion.

vein

normal

100%, and of 27).

successful of anasto-

was

visu-

hindered

was

96%

(26

Eight patients had massive gastric, splenic, coronary, or esophageal vances on MR angiograms (Figs 3, 4). Ascites was present on MR images in

nine

patients,

(>20

cm

massive

vertical

splenomegaly

span)

in five,

en-

barged gonadal veins in four (Fig 4), and distended paravertebral and azygous veins in one; two patients had spontaneous splenorenal shunts. Large umbilical veins were present in two patients (Fig 5). Ascites had variable ography,

angiography.

anatomy,

MPV

its specificity

signal with

intensity at MR angibright and dark cornfrequently detected in the

were visualized MR angiographic

at MR

underwent by means

por-

patient

by metal artifact from surgical clips. Thus, the sensitivity of MR angiography for showing MPV occlusion was

sion

MPV

intrahepatic In one

vascular

ponents

of the

third patient died to bypass the por-

with

of the

Technically satisfactory MR angiograms were obtained in all cases, and no studies were terminated prematurely. Eighteen patients had anatomically normal MPVs at MR with antegmade flow (Fig 1). Three patients had small (