393

Color Doppler Imaging of

Portosystemic Shunts

Edward G. Grant1 This study was designed to investigate the utility of color Doppler sonography in the
Franklin N. Tessler1 evaluation of portosystemic shunts. Thirty-one patients with a total of 32 shunts were
Antoinette S. Gomes1 imaged. The types of shunts examined included portacaval, five; mesocaval, eight;
distal splenorenal(Warren), 14; and mesoatrial, five. Sonography was performed without
Chonita L. Holmes1
knowledge of the status of the shunt, although the type of shunt was known before
Rita R. Perrella1
beginning the study. The sonographic studies were evaluated to determine their sensi-
Andre J. Dueninckx1
tivity and specificity on the basis of a prospective comparison with angiography or MR
Ronald W. Busuttil2 imaging (22 cases). The possible advantages of color Doppler over duplex Doppler
sonography in evaluating portosystemic shunts were also investigated, as was the
ability of color Doppler sonography to image specifically the shunt anastomoses. Color
Doppler sonography successfully inferred shunt patency (17 cases) or thrombosis (five
cases) in all 22 shunts for which correlative imaging was available (sensitivity = 100%,
specificity = 100%). In comparing duplex with color Doppler sonography in all 32 shunts,
the two techniques were almost equally effective in establishing patency in portacaval,
mesocaval, and mesoatrial shunts. Duplex Doppler sonography, however, provided
useful diagnostic information in only four of 14 splenorenal shunts. Color Doppler
correctly inferred patency or thrombosis in all 14. Among all 32 shunts, the anastomosis
was shown clearly by color Doppler in 23, probably in four, and not at all in five.
Our results suggest that color Doppler sonography is an excellent method for the
evaluation of all varieties of surgically created portosystemic shunts. In particular, color
Doppler sonography appears to be superior to duplex Doppler sonography in imaging
splenorenal communications.

AJR i54:393-397, February 1990

Portosystemic shunts are constructed to palliate symptomatic portal hyperten-

sion [i 2] or to provide decompression
, of the hepatic vasculatune in patients with
Budd-Chiani syndrome [3]. All such shunts may thrombose occasionally, though
the frequency varies with the type of shunt [4, 5]. Acute blockage of a portosystemic
shunt may result in a life-threatening recurrence of symptoms.
Angiography has been the definitive method of evaluating portosystemic shunts
for some time. However, angiography is invasive, and the anatomically isolated
portal system may be difficult to opacify [6]. Numerous imaging techniques,
including neal-time sonography [7], contrast-enhanced CT [8, 9], and MR imaging
[i 0], have been used to evaluate portosystemic shunts noninvasively. Recently,
Received July 1 7, 1 989; accepted after revision
duplex Doppler sonography has been used in this regard as well [9, 1 i -1 6]. The
September 19, 1989.
1 Department of Radiological Sciences, Univer- results of evaluations with duplex Doppler sonography have been encouraging, but
sity of California, Los Angeles, Schcol of Medicine,
highly dependent on the location of the shunt [9, 1 i i 5]. We undertook, this study
10833 LeConte Ave., Los Angeles, CA 90024. Ad- to assess the possible role of color Doppler imaging in the evaluation of surgically
dress reprint requests to E. G. Grant.
created portosystemic shunts.
2 Department of Surgery, University of California,
Los Angeles, Schcol of Medicine, Los Angeles, CA
90024.
Subjects and Methods

0361 -803X/90/1 542-0393 Thirty-two portosystemic shunts in 31 patients were evaluated with color and duplex
© American Roentgen Ray Society Doppler imaging. The ages of the patients at the time of sonographic examination ranged
394 GRANT ET AL. AJR:154, February 1990

from 1 7 to 70 years (mean, 41 years). Sixteen men and 1 5 women splenorenal shunts to be considered patent, appropriately directed
were evaluated. Fourteen of the 31 patients were referred to our flow had to be visualized in both limbs of the shunt (portal vein/
institution for possible liver transplantation. These 1 4 patients under- inferior vena cava for portacaval, splenic vein/renal vein for spleno-
went color and duplex Doppler imaging of the entire portal venous renal). For mesoatrial and mesocaval shunts, visualization of appro-
system as part of their preoperative evaluations. Fourteen patients priately directed flow in any portion of the synthetic graft was taken
were referred specifically for evaluation of shunt patency; three to imply patency. Sensitivity and specificity were determined by
additional shunts were identified on sonograms obtained for other comparing the results of color Doppler sonography with angiography
reasons. Fifteen of the 31 patients were symptomatic with ascites or and/or MR imaging. Angiography was performed 1 4 times and MR
recent gastroesophageal hemorrhage at the time the sonogram was imaging 1 1 times in a total of 22 patients. Angiography was performed
obtained. These 1 5 symptomatic patients were from both the pre- with standard technique. Selective celiac (and/or splenic) and superior
transplantation and nontransplantation groups. mesenteric artery injections were performed in all patients. In some
Underlying disease processes included chronic active hepatitis (10 cases the shunts were entered directly. Hepatic vein wedge pressures
patients), primary biliary cirrhosis (six patients), Budd-Chiari syndrome were obtained as indicated. MR imaging was performed with spin-
(six patients), alcoholic liver disease (six patients), sclerosing cholan- echo technique in oblique, axial, and coronal projections. Correlative
gitis (one patient), cystic fibrosis (one patient), and hemochromatosis imaging procedures were reviewed by one member of our group who
(one patient). The types of shunts encountered were varied: porta- had no knowledge of the results of the sonographic studies. The
caval (five), mesocaval (eight), distal splenorenal (1 4), and mesoatrial results of the sonograms were then compared with MR imaging and
(five). The time between construction of the shunt and the initial angiography in a blinded fashion. Sensitivity and specificity of color
sonographic evaluation ranged from 1 day to 9 years. The results of Doppler imaging were based on the results in these 22 shunts.
color imaging in our initial three patients have been reported else- The results of duplex vs color Doppler imaging were then compared
where [15]. in all 32 portosystemic shunts. Duplex Doppler examinations were
All Doppler studies were performed with a commercially available performed in the usual manner. The shunt was first located, if
unit (Ultramark 9, Advanced Technology Laboratories, Bothell, WA). possible, using real time. The Doppler cursor was then appropriately
Duplex and color Doppler imaging was performed with phased- and placed within the shunt and its patency was established on the basis
linear-array technology; 2.25-, 3.5-, and 5.0-MHz transducers were of receiving the expected Doppler signal. A duplex Doppler study
used. At the beginning of the examination, the patient’s chart was was considered unsuccessful if real-time visualization of the shunt
reviewed to determine the type of shunt present. The sonologist did (or the limbs of the shunt as described above for color Doppler
not know was considered
if the shunt patent on clinical grounds and imaging) was insufficient to allow adequate placement of the Doppler
had no knowledge of the results of other imaging procedures. The cursor.
expected area of the shunt was scanned first with duplex Doppler The ability of color Doppler imaging to depict shunt anastomoses
and then with color Doppler imaging. The sonographic studies were also was evaluated in all 32 shunts.
then evaluated to address three specific questions: (1 ) What are the
sensitivity and specificity of color Doppler sonography in the evalua- Results
tion of portosystemic shunts? (2) Is color superior to duplex Doppler
in assessing portosystemic shunts? (3) Can color Doppler imaging Angiography and/or MR imaging demonstrated shunt pa-
reliably depict shunt anastomoses? tency in 1 7 of 22 cases; five shunts were occluded. Color
The criteria for inferring shunt patency varied considerably depend- Doppler sonognaphy successfully demonstrated shunt pa-
ing on the type of shunt under investigation. For portacaval and tency in all 1 7 (Figs. i -4). Occlusion was also diagnosed

Fig. 1-Portacaval shunt. Fig. 2.-Mesoatrial shunt. Longitudinal color

A and B, Longitudinal (A ) and transverse (B) color Doppler images show communication between Doppler image shows flow within superficial, sub-
portal vein and inferior vena cava (IVC) (arrows). Considerable turbulence is common at shunt costal portion of mesoatrial shunt. Note brightly
anastomoses, as evidenced by mixture of colors in these areas. During portions of cardiac cycle, echogenic walls of synthetic shunt (arrows). Al-
blood may reflux into portal system in response to normal phasic flow reversals in ivc. Note red though flow in shunt is directed toward right
color in portal vein (B); flow is directed toward Doppler beam. atrium, color changes from red to blue as flow
advances and recedes in relation to Doppler beam.
AJR:154, February 1990 COLOR DOPPLER OF PORTOSYSTEMIC SHUNTS 395

correctly in the remaining five (Fig. 5). Specificity and sensitiv- plex Doppler sonography, however, fulfilled our criteria for a
ity, therefore, were 1 00%. These findings are summarized in successful study in only four of i 4 splenorenal shunts
Table 1 . In one of the 1 7 patients, a splenorenal shunt was (Table 2).
shown to be patent by both sonography and MR imaging. Among the 32 shunts, the anastomosis was imaged by
MR imaging, however, showed a narrowing of a small left color Doppler sonography cleanly in 23, probably in four, and
renal vein as it passed between the superior mesentenic artery not at all in five (Table 3).
and the aorta. At the time of the sonogram, the left renal vein
could not be followed as far as the aorta. While this shunt
was patent, its function may have been compromised. Both
Discussion
sonography and MR imaging showed the left renal vein to be Sonognaphy has been advocated since 1 977 as a nonin-
draining into large retnopenitoneal collatenals. This patient was vasive method of evaluating portosystemic shunts [6]. So-
not symptomatic at the time of the sonogram. nognaphy is particularly applicable in acutely ill on uncomfort-
In comparing color and duplex Doppler sonognaphy in all able patients. In addition, the inherent properties of sonogra-
32 shunts, we found that all portacaval and mesoatrial shunts phy make it an attractive screening procedure immediately
were adequately evaluated by both techniques. Duplex Dop- after placement of a shunt or before liven transplantation.
pIer was also successful in the evaluation of seven of eight Unfortunately, the use of neal-time sonography in the evalua-
mesocaval shunts (see Fig. 3B). Color Doppler sonography tion of portosystemic shunts has been limited, since its ana-
correctly assessed patency in all 1 4 splenonenal shunts. Du- tomically oriented images can only suggest patency. Duplex

Fig. 3.-Mesocaval shunt.

A, Longitudinal color Doppler image shows su-
perior mesenteric vein (SMV) anterior to synthetic
graft (arrow). Color signals are often impossible
to demonstrate throughout an entire graft in one
image because of changing relationship between
flow direction and Doppler beam.
B, Patency of mesocaval shunt in another pa-
tient is confirmed by showing flow entering distal
inferior vena cava (I) from another direction. With
persistent scanning, walls of shunt were eventu-
ally appreciated (curved arrow). Note posterior
acoustic shadowing from graft walls (straight ar-
rows).

Fig. 4.-Distal splenorenal shunt. Fig. 5.-Thrombosis of mesoatrial shunt is in-

A, Real-time image from duplex Doppler scan performed via a left anterior approach. No major dicated by complete absence of color within echo-
vessels are identified in expected area of splenic limb of Warren shunt. genic walls (arrows).
B, Color Doppler image from same location clearly shows flow in splenic vein (S) coursing toward
left renal vein. Anastomosis and left renal vein were imaged by scanning from other vantage points.
Duplex Doppler sonography is unsuccessful in these situations.
396 GRANT ET AL. AJR:154, February 1990

TABLE 1: Sensitivity and Specificity of Color Doppler Imaging Doppler imaged the flow within. Once the graft was located,
in the Evaluation of Portosystemic Shunts
either color or duplex Doppler sonognaphy could be used to
Angiography/MR Correctly Diagnosed establish patency. Of all shunts, Doppler signals were the
Type of Shunt by Color Doppler most difficult to elicit from within the synthetic walls of me-
Patent Thrombosed Total Imaging socaval grafts. In the single mesocaval shunt that was not
Portacaval 3 1 4 4 seen with duplex Doppler sonognaphy, the initial feature iden-
Mesoatrial 3 2 5 5 tified by colon Doppler sonognaphy was the presence of a
Mesocaval 5 1 6 6 localized area of reversed flow in the mid-distal inferior vena
Splenorenal 6 1 7 7
cava. As the inferior vena cava has no major infrarenal
Total 17 5 22 22 branches, such flow could only be from the patient’s shunt.
Duplex Doppler sonography was not capable of graphically
depicting this flow phenomenon and was unsuccessful in this
TABLE 2: Duplex vs Color Doppler Imaging in the Evaluation of case.
Portosystemic Shunts While duplex Doppler imaging has been considered as a
potential screening procedure for all types of portosystemic
Adequate Shunt
Total shunts, a review of the literature reveals that most of the
Type of Shunt Visualization
patients studied had portacaval communications [i i-is].
. Duplex Color Unfortunately, portacaval shunts are being constructed less
Portacaval 5 5 5 frequently as liver transplantation becomes an accepted form
Mesoatnial 5 5 5 of therapy for patients with end-stage liven disease and portal
Mesocaval 8 7 8 hypertension. Mesocaval and splenonenal shunts, therefore,
Splenorenal 14 4 14
are being performed more frequently. Although duplex Dop-
Total 32 21 32 pIer sonography has been relatively successful with mesoca-
val shunts, the evaluation of distal splenonenal shunts has
been problematic. Foley et al. [9], in fact, specifically ad-
TABLE 3: Ability of Color Doppler to Image Shunt Anastomoses dressed left upper quadrant shunts and found duplex Doppler
sonognaphy thoroughly inadequate. Our results with duplex
Total Not Doppler sonography, though somewhat better than those of
Type of Shunt Questionable
No. Imaged Imaged
Foley et al., were still disappointing. The venous limbs of
Portacaval 5 4 1 0 splenonenal shunts were visualized by real time in only four
Mesoatrial 5 2 3 0
of i 4 cases. Adequate placement of a Doppler cursor, there-
Mesocaval 8 8 0 0
Splenorenal 14 9 1 4 fore, was not possible. Color Doppler sonognaphy, on the
other hand, was fan more successful in the left upper quad-
Total 32 23 5 4
rant. The color Doppler image was visible through a relative
veil of bowel gas and backscatten. Adequate visualization of
the splenic and renal limbs of Warren shunts was possible in
Doppler sonognaphy, however, provides physiologic flow in- all of our patients. While the thin walls of native veins are not
formation and has been used successfully by a number of brightly echogenic, they apparently offer little impedance to
authors in the evaluation of portosystemic shunts [9, 11- the transmission of the Doppler beam. The implanted splenic
1 6]. Colon imaging is the latest refinement of sonographic vein is easily located with colon Doppler imaging.
technology, and its potential in the evaluation of portosys- While color Doppler imaging was capable of inferring shunt
temic shunts has not, to our knowledge, been investigated. patency in all patients in our series, the shunt anastomosis
Colon Doppler sonognaphy was capable of directly imaging was somewhat more challenging to find. As expected, visibil-
shunt flow and inferring patency vs thrombosis in all patients. ity varied with the type of shunt being evaluated. The anas-
Sensitivity and specificity were 1 00% in our series. tomosis between the portal vein and the inferior vena cava
Previous studies of portosystemic shunts using duplex was seen readily in almost all patients. The liver provided an
Doppler sonography have found the technique to be relatively excellent acoustic window. Likewise, the junction between
accurate. The success of duplex Doppler sonognaphy, how- the synthetic graft and the inferior vena cava in mesocaval
ever, is largely dependent upon the location of the shunt. shunts was readily imaged once an appropriate acoustic
Duplex Doppler sonography can image portacaval and me- window was found. Demonstration of the anastomoses in
soatnial shunts quite adequately. Our study corroborates ear- mesoatrial and splenonenal shunts, however, was more com-
lien work in this regard [i 0, i 2-i 4]. Although colon Doppler plicated. In mesoatrial shunts, the junction between the su-
sonognaphy increased confidence and markedly lessened perion mesenteric vein and the graft lies deep within the
scanning time, it offered no true diagnostic advantage oven abdomen. This region is notoriously difficult to evaluate with
duplex Doppler in imaging these two types of shunts. Our sonognaphy because of surrounding bowel gas and mesen-
results also indicate that mesocaval shunts are most often tenic fat. This situation is probably further worsened by the
adequately imaged by duplex Doppler sonography. In our high attenuation of the synthetic graft walls. Fortunately, long
experience, real-time scanning, in fact, typically identified the portions of all mesoatrial shunts are easily imaged as they
echogenic walls of the synthetic graft more readily than color pass cephalad and come to lie directly beneath the anterior
AJR:154, February 1990 COLOR DOPPLER OF PORTOSYSTEMIC SHUNTS 397

abdominal wall. In our experience, confirmation of flow within scanning offers diagnostic information in most patients with
even a small portion of a mesoatnial (or mesocaval) shunt is portacaval, mesocaval, and mesoatrial shunts, colon Doppler
sufficient to establish patency. imaging is performed more rapidly, and anatomic delineation
The anastomosis between the splenic vein and left renal of shunt physiology increases physician confidence. In the
vein in Warren shunts also may be difficult to image. The evaluation of splenonenal shunts, however, our study shows
splenic limb of the shunt usually can be followed dorsally for color Doppler sonognaphy to be superior to duplex Doppler
a considerable distance when scanning from a left subcostal sonography and the only adequate sonognaphic method of
approach. Left renal vein flow, however, is normally directed evaluation.
perpendicular to the splenic vein. Doppler signals from the
left renal vein, therefore, are best received by scanning
REFERENCES
through the left flank. Because imaging of the splenic and
renal veins is optimally performed from two different vantage 1 . Malt AA. Portasystemic venous shunts. Part 1 . N EngI J Med 1976;295:
points, the anastomosis may be difficult to image in a single 24-29
2. Malt AA. Portasystemic venous shunts. Part 2. N Eng! J Med 1976;295:
section. With experience and meticulous technique, colon
80-86
Doppler sonography should be capable of visualizing the 3. Cameron JL, Maddrey WC. Mesoatrial shunt: a new treatment for the
anastomosis of patent splenonenal shunts in most patients. Budd-Chiari syndrome. Ann Surg 1978;187:402-406
We have found that scanning the patient when the stomach 4. Schwartz SI, ed. Principles of surgery, 3d ed. New York: McGraw-Hill,
1979: 1297-1 315
is empty is particularly helpful.
5. Cameron JL, Kadir 5, Pierce WS. Mesoatrial shunt: a prosthesis modifi-
In our patient with a splenorenal shunt and angiognaphically cation. Surgery 1984;96:114-116
proved thrombosis, the actual site of occlusion was not seen. 6. Abrahms HL, ed. Angiography, 2d ed. Boston: Little, Brown, 1971:1073-
This is to be expected in splenorenal shunts since the walls 1078
of native veins are not sufficiently echogenic to outline the 7. Goldberg BB, Patel J. tJtrasonic evaluation of portacaval shunts. JCU
1977;5:304-306
shunt if a color signal is absent. Thrombosis, however, was
8. Gleysteen JJ, Foley WD, Lawson TL, Unger GF. Patency evaluation of
inferred when a tangle of vessels was found extending from distal splenorenal shunt with dynamic computed tomography. Surg Gyne-
the spleen to large retropenitoneal collaterals. Among these col Obstet i982;154:689-694
vessels, the splenic limb of the shunt could not be defined as 9. Foley WD, Gleysteen JJ, Lawson TL, et al. Dynamic computed tomography
and pulsed Doppler sonography in the evaluation of spienorenal shunt
an individual structure. The presence of shunt dysfunction in
patency. J Comput Assist Tomogr 1983;7: 106-112
this patient was particularly obvious since a routine postop- 10. Bemardino ME, Steinberg HV, Pearson TC, Gedgaudas-McCIees AK,
enative study 6 weeks earlier showed the typical single vessel Torres WE, Henderson JM. Shunts for portal hypertension: MR and
extending from the splenic hilum to the left renal vein. No angiography for determination of patency. Radiology 1986;158:57-61
collatenals were present at the time of the earlier study. The 11. Ackroyd N, Gill A, Griffiths K, Kossoff G, Reeve T. Duplex scanning of the
portal vein and portasystemic shunts. Surgery 1986;99:591-597
appearance of collatenals in the patient with thrombosis was
12. Finn JP, Gibson AN, Dunn GD. Duplex ultrasound in the evaluation of
similar to that in the patient with suspected shunt dysfunction portacaval shunts. Clln Radio! 1987;38:87-89
secondary to narrowing of the central left renal vein. On the 13. Forsberg L, Holmin T. Pulsed Doppler and B-mode ultrasound features of
basis of these two cases, abnormal shunt physiology (or interposition meso-caval and porta-caval shunts. Acta Radio! (Diagn]
(Stockh) i983;24:353-357
actual thrombosis) should be suspected in any patient with a
14. Lafortune M, Patnquin H, Pomier G, et al. Hemodynamic changes in portal
splenorenal shunt in whom left upper quadrant collatenals are circulation after portosystemic shunts: use of duplex sonography in 43
identified instead of a well-defined splenic limb. patients. AJR i987;149:701-706
Our study has found color Doppler sonography to be an 15. Patriquin H, Lafortune M, Weber A, Blanchard H, Garel L, Roy C. Surgical
excellent noninvasive method of imaging portosystemic portosystemic shunts in children: assessment with duplex Doppler US.
Radio!ogy 1987;165:25-28
shunts of all varieties. The examination is well tolerated and 16. Grant EG, Perreila A, Tessler FN, Lois J, Busuttil A. Budd-Chiari syndrome:
may even be performed portably, if necessary. While duplex the results of duplex and color Doppler imaging. AJR i989;1 52:377-381