- Email: [email protected]
Portal hemodynamics in patients with gastric varices
GASTROENTEROLOGY
Portal Hemodynamics Gastric Varices
in Patients With
A Study in 230 Patients With Esophageal Gastric Varices Using Portal Vein Catheterization KIYOSHI MASAO
WATANABE, OHTO,
KUNIO
and KUNIO
1988:95:434-40
KIMURA,
SHOICHI
and/or
MATSUTANI,
OKUDA
First Department of Medicine, Chiba University School of Medicine, Chiba, Japan
The hemodynamic features of gastric varices are not well documented. The purpose of this study was to investigate the nature of hepatofugal collateral veins, their origins, the direction of blood flow in the major veins and collateral veins, and portal venous pressure. To this end, 230 patients, mostly cirrhotic, who had esophageal or gastric varices, or both, demonstrated by endoscopy were investigated by portal vein catheterization. The findings were correlated with endoscopically assessed degrees of varices. Gastric varices were seen in 57% of the patients with varices due to portal hypertension. In most of the patients with advanced gastric varices, esophageal varices were minimal or absent. When patients with gastric varices were compared with those having predominantly esophageal varices, it was found that advanced gastric varices were more frequently supplied by the short and posterior gastric veins, they were almost always associated with large gastrorenal shunts, and portal venous pressure in patients with large gastric varices was lower. Chronic portal systemic encephalopathy was more common in patients with large gastric varices due to hepatofugal flow of superior mesenteric venous blood in the splenic vein than in patients with predominantly esophageal varices. Thus, the hemodynamics in patients with large gastric varices are distinctly different from those in patients with mainly esophageal varices, and such differences seem to account for the differing incidence of chronic encephalopathy and variceal bleeding. sophageal varices are the most common complication of portal hypertension that has serious consequences. Although gastric varices are less common and bleed less frequently (l-3), they are just as
E
important as esophageal varices in terms of prognosis. Despite the vast clinical experience gained in the past with gastric varices, little is known about the hemodynamic features of such patients. In this study, 230 patients with portal hypertension who were found by endoscopy to have varices in the esophagus or stomach, or both, were investigated by portal vein catheterization. The hemodynamic features of esophageal and gastric varices were compared in relation to the size of varices, portal venous pressure, portal systemic encephalopathy, and renal shunts.
Patients
and Methods
During the period from September 1980 to April 1984, 230 patients (175 men, 55 women, aged 29-79 yr) who were endoscopically found to have either esophageal varices, gastric varices, or both, were investigated by portal vein catheterization (4-6) in the First Department of Medicine, Chiba University Hospital and its affiliated hospitals. The hemodynamic study was done in these patients to assess whether a surgical procedure was indicated and, if so, which operation was preferred. The study group comprised 201 cases of liver cirrhosis, including 32 cases with hepatocellular carcinoma smaller than 5 cm with no intraportal invasion (studied by ultrasound), 20 cases of idiopathic portal hypertension (7,8), 4 cases of primary biliary cirrhosis, and 5 cases of miscellaneous diseases. The clinical status of the patients was Child’s A in 86 cases, Child’s B in 111 cases, and Child’s C in 33 cases not having ascites. Although 79 patients (72 men and 7 wom-
Abbreviations used in this paper: EV, esophageal varices; GR, gastrorenal; GV, gastric varices; PVP, portal venous pressure; SMV, superior mesenteric vein. 0 1988 by the American Gastroenterological Association 0016-5085/88/$3.50
HEMODYNAMICS IN GASTRIC VARICES
August 1988
Figure
1. Endoscopic appearance of GV. A. Mild varices of grade GV (+) seen in the cardia. B. Medium-sized running posteriorly from the cardia. C. Marked varices of grade GV (3+) looking like a submucosal
en) had a drinking habit, only 35 were assumed to have had alcoholic cirrhosis and the remainder most likely had posthepatitic cirrhosis as judged from the course of serum aminotransferase levels. Informed written consent was obtained from each patient before portal catheterization. The patients who had had an operation for portal hypertension were not included. There were 12 patients with chronic portal systemic encephalopathy characterized by episodic attacks of encephalopathy who had otherwise relatively well-maintained liver function tests (9-11). Endoscopy The endoscopic findings of esophageal varices were classified according to the General Rules for Recording Endoscopic Findings set by the Japanese Research Society for Portal Hypertension (12), in which the form of varices is graded as F, (straight), F, (winding), and F3 (nodule-beaded). The degree of gastric varices (GV) was classified into three grades according to Yasumoto (13): GV (+), mild varices in the cardia; GV (2+), winding veins seen in the cardia-fornix region; GV (3+), large varices in the cardia-fornix looking like a tumor (Figure 1). Percutaneous
Transhepatic
varices of grade GV (2+) tumor.
vein(s) that arise in the midportion of the splenic vein and run toward the cardia are called “posterior gastric,” and the veins that arise to the right of the midportion of the splenic vein and run toward the cardia are called “left or right gastric.” They used to be called the “coronary” veins. The collaterals that connect these gastric veins and the left renal vein studied on the venograms are called gastrorenal (GR) shunts. They were graded according to the diameter of the left suprarenal vein immediately before entering the left renal vein as follows: +, ~5 mm in diameter; 2+, 5-10 mm: 31, >lO mm (Figure 2). Similarly, splenorenal shunts were defined as the connections between the splenic vein and the left renal vein, and they were graded according to the diameter of the connecting vein shortly before it entered the left renal vein as follows f. x:5 mm; 2+, 5-10 mm: 3+, >lO mm. The flow direction of the contrast medium in these veins and collaterals was also studied. For the statistical analysis, either the ,$ test or the Student’s t-test (comparison of averages) was used.
Results Endoscopic
Catheterization
The ultrasound-guided procedure already detailed (6,14,15) was used. After measurement of portal venous pressure with the zero point at 10 cm above the table, contrast medium was injected through the catheter tip into the splenic hilum and into the superior mesenteric vein for the study of collateral veins and blood flow direction. By these venographic procedures, the routes of hepatofugal blood flow that formed varices were clarified. The sizes of collateral veins were measured on the films to determine the major route of portal-systemic shunting. In this study, the short gastric veins are defined as those veins that arise in the splenic hilum or within the spleen and course from the greater curvature of the stomach around the fornix. The
435
The five groups gastric
patients
were
depending
varices
assessment
Findings
(EV
divided
into the following
on the size of esophageal and
of the degree
GV,
respectively).
and In the
of EV and GV, F1 was taken
as equivalent to GV (+), F2 equivalent to GV (2+), and F, equivalent to GV (3+). Thus, the five groups were EV: EV > GV: EV + GV:
there were only EV (n = 99, 43%), EV were larger than GV (n = 98, 42.5%), EV were about the same in degree as GV (n = 10, 4.3%),
436
Figure
WATANABE
ET AL.
GASTROENTEROLOGY
2. Portographic features of three grades of GR shunt. A. Gastrorenal shunt (+). in which the diameter of the suprarenal vein entering the left renal vein is <5 mm. B. Gastrorenal shunt (2+), the suprarenal vein is between 5 and 10 mm. C. Gastrorenal shunt (3+), the suprarenal vein is >lO mm. C’. Late-phase film of C. GV. gastric varices; IVC, inferior vena cava: LRV, left renal vein; large arrow, the inferior diaphragmatic vein; two small orrows. the left suprarenal vein.
EV < GV: GV:
GV were larger than EV (n = 15, 6.5%), there were only GV (n = 8, 3.5%).
Thus, 131 of 230 patients degree. Venographic
(57%)
had GV of varying
Findings
The venographic findings in these 230 patients were divided into three groups: the right type (R > L), in which the left/right gastric vein was the predominant collateral route to form the varices; the left type (R < L), in which posterior/short gastric veins were the major route that formed the varices; and the equal type (R + L), in which the left/right R>L
R%L
gastric vein and the posterior/short gastric veins contributed about equally to the formation of varices (Figure 3). The majority, 179 cases, showed the right type (77.8%), 20 cases were of the equal type (8.7%), and 31 cases were of the left type (13.5%). As shown in Table 1, the majority of EV and EV > GV cases had the right type (83% and 86%, respectively]. By contrast, only 25% of GV cases and 33% of EV < GV cases had the right type, about one-half of GV and EV < GV cases showed the left type, and the remainder had the equal type.
Table
1. Endoscopic Grading of Esophageal and Gastric Varices in Relation to the Collateral Veins That Were Supplying Them
R
Figure
Vol. 95, No. 2
3. Three types of venous supply to gastric varices as classified by portography. R > L. right type; R + L, equal type; R < L, left type. EV, esophageal varices; GV, gastric varices; PV, portal vein; SV, splenic vein: L, posterior and/or short gastric vein; R. left or right gastric vein.
EV EV > GV EV + GV EV< GV GV Total
Types R>L 83 84 5 5 2
(83%) (86%) (50%) (33%) (25%)
179 (77.8%)
of venous
supply R
R+L 5 8 3 2 2
(5%) (8%) (30%) (13%) (25%)
20 (8.7%)
to varices
11 6 2 8 4
(11%) (6%) (20%) (53%) (50%)
31 (13.5%)
Total 99 98 10 15 8
(100%) (100%) (100%) (100%) (100%)
230 (100%)
EV. esophageal varices; GV, gastric varices; R. left or right vein: L, posterior or short gastric veins, or both.
gastric
HEMODYNAMICS
August
1988
Table
2. Relationship Between Grading of Esophageal and Gastric Varices and the Size of Gastrorenal Shunts Grade
Type
of varices
-
Frequency of GR shunt
+
2+
3+ 1
17 (17%) 21 (21%)
82
14
2
EV > GV (n = 981
77
14
4
3
4
1
1
4
6 (60%)
3
0
4
8
12 (80%)
0
0
1
7
8 (100%)
166
29
12
23
64 (27.8%
GV(n = 10) EV < GV (n = 15) GV (n = 6)
Total EV, esophageal
varices;
GV. gastric
Relationship Between Degree of Gastrorenal
varices;
Types Shunt
Relationship Between Types of Varices Portal Venous Pressure and Gastrorenal Shunts
and
and
PVP” (mm&O)
EV (n = 991 EV > GV (n = 981 EV + GV (n = lo] EV i GV (n = 15) GV (n = 8) EV, esophageal varices: GV, gastric pressure. ‘Mean 2 SD. ’p < 0.01.
326 t 66b 340 k 65 337 + 63 302 + 75 240 + 37b
varices;
PVP, portal
_
+
Yes
339 + 66”
343 ? 65
(n = 84)
(n f 15)
3+ 277 C 67”
131
2 -t 345 -t 56 ,
(n 7 10)
(11 = 22)
venous
77”.
(n = 4’7) No
3. Relationship Between Grading of Esophageal and Gastric Varices and Portal Venous Pressure of varices
GV
Total of cases
of GR shunt
311 t
As summarized in Table 3, the average portal venous pressure (PVP) was 326 + 66 mmHzO (mean ? SD] in the EV group, 340 t 65 mmHzO in the EV > GV group, 337 ‘- 63 mmHzO in the EV + GV group, 302 rt 75 mmHzO in the EV < GV group, and 240 +- 37 mmHsO in the GV group. The difference between the EV and GV groups was significant (p < 0.01). Among the cases with GV, the correlation between the grade of GR shunt and PVP was analyzed (Table 4). When 84 patient.s not having GR shunts and 47 having GR shunts were compared for PVP, the latter (311 -+ 73 mmHzO) had a significantly lower PVP compared with the former (339 +- 66 mmHzO) (p < 0.05). The portal venous pressure was
Type
437
Between the Degree of Shunt and Portal Venous Pressure
of total)
CR, gastrorenal
of Varices
VARICES
343 k 58b
Table 2 illustrates this relationship. Whereas only 17% of EV (EV only) cases and 21% of EV > GV cases had GR shunts, 80% of EV < GV cases and 100% of GV cases had GR shunts (EV and EV > GV combined vs. EV < GV and GV combined, p < 0.001). In these latter groups, the degree of GR shunt was also marked in the majority.
Table
4. Relationship Gastrorenal
Degree
of GR shunt
EV (n = 99)
EV+
Table
IN GASTKIC
326 + 66’.
324 + 59
(n = 82)
(n = 17)
GR, gastrorenal; 0.05 (differences bols).
99
GV, gastric varices. ” p < 0.05; ” p i 0.01: ’ p
in 22 patients who had GV and ‘_ 67 mmHzO (3s) GR shunts; it was significantly lower than that in any other group, even than that of the (+) and (2-t) groups combined (p < 0.01). Thus, it seemed that if GV became significantly large, PVP would decrease. No such differences were noted among groups without GV with regard to the size of GR shunts. Altogether there were 21 patients who had SR shunts in the whole series: 12 had (+) shunts, 5 (2+) shunts, and 4 (3+) shunts. The portal venous pressure in these groups was 328 + 67. 337 k 32, and 313 + 111 mmH,O, respectively. 277
Direction Varices
of Blood
Flow
in Relation
to
The blood flow from the superior mesenteric vein (SMV) into the portal or splenic vein, or both, could be divided into three types: A, most SMV blood flowed into the portal vein. opacifying the intrahepatic portal branches clearly: B, part of the SMV blood flowed into the splenic (and left gastric] vein with a resultant reduced flow into the portal vein and poor opacification of intrahepatic portal branches; and C, all SMV blood flowed into the splenic and left gastric veins without opacification of the portal vein and its tributaries, as illustrated in Figure 4. Table 5 correlates the five types of varices and the three blood flow types. Type A flow was the predominant type in patients with EV only (EV) or mainly esophageal varices (EV > GV). As GVs became large (groups EV + GV, EV < GV. and GV), there were more type B flows (40%, 33% and 63%, respectively); 2 of 8 (25%) cases in the GV group had total diversion of SMV flow away from the liver (type C). In 7 of 8 cases (88%), the flow in the splenic vein near its junction with the SMV was retrograde.
438
WATANABE
ET AL.
GASTROENTEROLOGY
Type A
Type B
I
Table
Type C
I
6. Frequency in Relation
of Portal Systemic Encephalopathy to Grading of Esophageal and Gastric Varices
Direction of portal venous Type of varices I
Visualization of IHPB Figure
good
poor
“0
4. Schematic representation of the three types of portal hemodynamics with respect to the flow of superior mesenteric venous blood. IHPB, intrahepatic portal branches.
Relationship of Varices and Chronic Portal Systemic Encephalopathy Of the 230 patients studied, 12 had a typical (9-11). Five chronic portal-systemic encephalopathy of the 12 encephalopathic patients had marked GV (41.7%). Therefore, the relationship between varices and encephalopathy was investigated. As shown in Table 6, chronic encephalopathy was uncommon among patients of the EV and EV > GV groups [3.0% and 4.l%, respectively], whereas patients in the EV < GV and GV groups had portal systemic encephalopathy more frequently (12.0% and 25.0%, respectively). The difference between the EV and GV groups was significant (p < 0.011, and similarly there was a significant difference between the EV and EV > GV groups combined and the GV and EV < GV groups combined (p < 0.05). Encephalopathy was also correlated with the type of SMV blood flow. It occurred in 66.7% of the type C group, 30.8% of the type B group, and 1% of the type A group [Table 7).
Discussion In the barium swallow examination of the stomach, demonstration of GV needs certain uncom-
Table
5. Relationship
Between Grading of Esophageal and Gastric Varices and Types of Portal Hemodynamic Changes Shown in Figure 4 Type of hemodynamic
Type of varices EV EV > GV EV + GV EV < GV GV
B
A 95 89 6 10 1
(96%) (90%) (60%) (67%) (12%)
changes”
3 9 4 5 5
(3%) (10%) (40%) (33%) (63%)
1 0 0 0 2
C
Total
(1%) (0%) (0%) (0%) (25%)
99 98 10 15 8
EV, esophageal varices; GV, gastric varices. ’ A, good opacification of intrahepatic portal branches: B, poor opacification of intrahepatic portal branches; C, no opacification of intrahepatic portal branches (total hepatofugal flow of superior mesenteric venous blood).
Vol. 95, No. 2
EV EV EV GV
(n = 99) > GV (n = 98) 5 GV (n = 25) (n = 8)
Frequency of portal systemic encephalopathy 3 (3.0%)” 4 (4.1%) 3 (12.0%) 2 (25.0%)”
EV, esophageal varices; GV, gastric varices. ’ p < 0.01.
mon projections, without which they are often missed (13,161. In this respect, endoscopy is more diagnostic than the routine barium examination. With a second barium examination specifically aimed at detection of GV, Yasumoto (13) demonstrated GV in 55.2% of patients with cirrhosis, 63.2% of patients with cirrhosis and hepatocellular carcinoma, and 90.9% of patients with idiopathic portal hypertension. In the present study, the incidence of GV seen by the endoscope was 57%, which is close to the figures obtained by Yasumoto with the second barium study. However, in a study by one of the current authors (17), endoscopy failed to demonstrate GV in 9 of 46 patients with roentgenologically demonstrable GV. Perhaps neither procedure is perfect in the detection of small GV that may not be clearly distinguishable from normal rugae by endoscopy. The hemodynamic study of these patients with GV clearly showed that GV are frequently supplied from the short and posterior gastric veins in contrast to EV, which are known to be supplied mainly from the coronary (left and right gastric) vein. As the connection between GV and the portal vein through the short and posterior gastric veins is short, therapeutic sclerotherapy carries a risk of developing portal thrombosis. In patients with large GR shunts, a sclerosing agent could even go into the renal shunt, Table
7. Frequency of Portal Systemic Encephalopathy in Relation to the Type of Portal Hemodynamic Changes Type of portal hemodvnamic changes”
Frequency of portal systemic enceuhalouathv
A (n = 201) B (n = 26) C (n = 3)
2 (1.0%) 8 (30.8%) 2 (66.7%)
aA, good
opacification of intrahepatic portal branches; B, poor opacification of intrahepatic portal branches; C, no opacification of intrahepatic portal branches due to total diversion of superior mesenteric venous blood away from the liver (see Figure 4).
August
1988
and from there into the inferior vena cava. In view of the lower incidence of rupture, no prophylactic sclerosis seems warranted for gastric GV. They are covered by a relatively thick gastric mucosa as compared with EV, and rupture is uncommon unless a peptic ulcer develops immediately above them. What causes the major portal-systemic collateral route to develop as PVP increases? It could be purely anatomic, depending on the individual venous vascular makeup, or be due to a preexisting anomalous vein. The size of the potential collateral vein and the length of the connection by that vein between the portal axis vein and the systemic (caval) veins will also be a factor. At the moment, there is no way of predicting the major route of shunting in the early stage of portal hypertension, hence artificially altering the route to prevent future variceal bleeding. The so-called red color sign (12), which portends future rupture and which is frequently seen over the surface of EV, was not observed in GV. Whereas the former involve veins in the lamina propria and mucosa (18-20).the same does not seem to occur in the latter. Perhaps due to the vascular anatomy in the cardia of the stomach, the short and posterior gastric veins connect more often with the left renal vein than does the left gastric vein, to form large shunts through the inferior diaphragmatic and the suprarenal vein. The portal venous pressure was significantly lower in patients with large GV and in patients with large GR shunts. There are occasional reports on normal portal pressure in the presence of a huge shunt (21) and reduction of PVP with enlargement of the renal shunt (22).Ohnishi et al. (23)studied the relationship between PVP and the degree of shunts in 52 patients with portal hypertension and found a correlation of a quadratic linear regression. In other words, PVP tended to decline as shunting became more extensive. The reason for the lower PVP observed in our patients with large GV and large GR shunts may be the same. Failure to demonstrate the same trend in patients with large splenorenal shunts was perhaps due to the small number of patients. Chronic portal systemic encephalopathy is a clinical syndrome in which the patient not severely debilitated by chronic liver disease develops encephalopathy in episodic bouts (g-11). It is commonly associated with large shunts, more often through the left renal vein than through other collateral routes (11,24-26).This was also observed in the current study with 12 patients. In such patients, the whole or part of the blood going out of the SMV flowed retrogradely into the splenic vein. Nitrogenous substances absorbed from the gut into the SMV enter the systemic circulation without going through, hence not being absorbed by, the liver in
HEMODYNAMICS
IN GASTRIC
VARICES
439
such patients. Thus, this syndrome should be suspected in patients with large GV or large renal shunts that can be seen by current real-time ultrasonograph (27).In this series, only 3% of the patients who had only EV had chronic portal-systemic encephalopathy. These data indicate that (a) GV occur nearly as frequently as EV but large GV are rather infrequent, (b) GV are different from EV in the route of blood supply, and (c) large GV are often associated with large GR shunts and chronic portal-systemic encephalopathy.
References M, Feldman M Jr. Gastric varices. Gastroenterology 1. Feldman 1956;30:318-21. 1953;60:46-50. 2. Evans JA, Delany F. Gastric varices. Radiology importance of gastric varices. N 3. Karr S. Wohl GT. Clinical Engl J Med 1960:263:665-9. A. Vang J. Transhepatic catheterization and 4. Lunderquist obliteration of the coronary vein in patients with portal hypertension and esophageal varices. N Engl J Med 1974;291: 646-9. K. Suzuki K. Musha H. Arimizu N. Percutaneous 5. Okuda transhepatic catheterization of the portal vein for the study of portal hemodynamics and shunts: a preliminary report. Gastroenterology 1977:73:279-84. K. Percutaneous transhepatic fi.Okuda K, Kimura K, Takayasu portography and sclerotherapy. Semin Liver Dis 1982;Z: 5 7-66. K, Omata M, eds. Idiopathic portal hypertension. 7. Okuda Tokyo: University of Tokyo Press, 1983. 8. Okuda K. Kono K, Onhishi K, et al. Clinical study of 86 cases of idiopathic portal hypertension and comparison with cirrhosis with splenomegaly. Gastroenterology 1984;86:600-10. 9. Sherlock S, Summerskill WHJ, White LP. Portal-systemic encephalopathy-neurological complications of liver disease. Lancet 1954;ii:453-7. 10. McDermott WV Jr, Adams R. Episodic stupor associated with an Eck fistula in the human with particular reference to the metabolism of ammonia. J Clin Invest 1954;33:1-9. 11. Summerskill WHJ, Davidson EA. Sherlock S. Steiner RF. The neuropsychiatric syndrome associated with hepatic cirrhosis and an extensive collateral circulation. Q J Med 1956;25: 245-66. 12. Japanese Society for Portal Hypertension. The general rules for recording endoscopic findings on esophageal varices. Jpn J Surg 1980;10:84-7. 13. Yasumoto M. Clinical observations on 100 cases of gastric varices. Jpn J Gastroenterol 1971:68:721-39. 14. Kimura K. Ohto M, Tsuchiya Y. et al. Single puncture method for percutaneous transhepatic portography using a thin needle. Radiology 1981;139:748-9. 15. Ohto M, Karasawa H, Tsuchiya Y. et al. [Jltrasonically guided percutaneous contrast medium injection and aspiration using a real-time puncture transducer. Radiology 1980:136:171-6. 16. Marshall JP II, Smith PD. Hoyumpa AM Jr. Gastric varices. Problem in diagnosis. Dig Dis 1977:22:947-55. 17. Okuda K, Yasumoto M, Goto A. Kunisaki T. Endoscopic observations of gastric varices. Am 1 Gastroenterol 1973;60: 357-65. 18. Kegaries DL. The venous plexus of the esophagus. Surg Gynecol Obstet 1934;58:46-51.
440
WATANABE
ET AL.
19. Butler H. The veins of the esophagus. Thorax 1951;6:276-96. 20. Noda T. Angioarchitectural study of esophageal varices. Virchows Arch [A] 1984;404:381-92. 21. Takashi M, Igarashi M, Hino S, Goto N, Okuda K. Chronic portal-systemic encephalopathy with normal portal vein pressure possibly due to noncirrhotic portal fibrosis. Dig Dis Sci 1984;29:669-73. 22. Ohnishi K, Nakayama T, Nomura F, Koen H, Okuda K. Reduction of portal vein pressure with the enlargement of portal-systemic shunts: observations made in one patient. J Clin Gastroenterol 1984;6:447-51. 23. Ohnishi K, Nakayama T, Koen H, et al. Interrelationship between type of spontaneous portal systemic shunt and portal vein pressure in patients with liver disease. Am J Gastroenterol 1985;80:5614. 24. Takashi M, Igarashi M, Hino S, et al. Portal hemodynamics in chronic portal-systemic encephalopathy. Angiographic study in seven cases. J Hepatol 1985;1:467-76.
GASTROENTEROLOGY
Vol. 95, No. 2
25.
Ohnishi K, Sato S, Saito M, et al. Clinical and portal hemodynamic features in cirrhotic patients having a large spontaneous splenorenal and/or gastrorenal shunt. Am J Gastroenterof 1986;81:450-5. 26. Lam KC, Juttner HU, Reynolds TB. Spontaneous portosystemic shunt-relationship to spontaneous encephalopathy and gastrointestinal hemorrhage. Dig Dis Sci 1981:26:34652. 27. Takayasu K, Moriyama N, Shima Y. et al. Sonographic detection of large spontaneous spleno-renal shunts and its clinical significance. Br J Radio1 1984;57:565-70.
Received July 13, 1987. Accepted March 9, 1988. Address requests for reprints to: Kunio Okuda, M.D., Ph.D., First Department of Medicine, Chiba University Hospital, Chiba, Japan (280).
Portal Hemodynamics Gastric Varices
in Patients With
A Study in 230 Patients With Esophageal Gastric Varices Using Portal Vein Catheterization KIYOSHI MASAO
WATANABE, OHTO,
KUNIO
and KUNIO
1988:95:434-40
KIMURA,
SHOICHI
and/or
MATSUTANI,
OKUDA
First Department of Medicine, Chiba University School of Medicine, Chiba, Japan
The hemodynamic features of gastric varices are not well documented. The purpose of this study was to investigate the nature of hepatofugal collateral veins, their origins, the direction of blood flow in the major veins and collateral veins, and portal venous pressure. To this end, 230 patients, mostly cirrhotic, who had esophageal or gastric varices, or both, demonstrated by endoscopy were investigated by portal vein catheterization. The findings were correlated with endoscopically assessed degrees of varices. Gastric varices were seen in 57% of the patients with varices due to portal hypertension. In most of the patients with advanced gastric varices, esophageal varices were minimal or absent. When patients with gastric varices were compared with those having predominantly esophageal varices, it was found that advanced gastric varices were more frequently supplied by the short and posterior gastric veins, they were almost always associated with large gastrorenal shunts, and portal venous pressure in patients with large gastric varices was lower. Chronic portal systemic encephalopathy was more common in patients with large gastric varices due to hepatofugal flow of superior mesenteric venous blood in the splenic vein than in patients with predominantly esophageal varices. Thus, the hemodynamics in patients with large gastric varices are distinctly different from those in patients with mainly esophageal varices, and such differences seem to account for the differing incidence of chronic encephalopathy and variceal bleeding. sophageal varices are the most common complication of portal hypertension that has serious consequences. Although gastric varices are less common and bleed less frequently (l-3), they are just as
E
important as esophageal varices in terms of prognosis. Despite the vast clinical experience gained in the past with gastric varices, little is known about the hemodynamic features of such patients. In this study, 230 patients with portal hypertension who were found by endoscopy to have varices in the esophagus or stomach, or both, were investigated by portal vein catheterization. The hemodynamic features of esophageal and gastric varices were compared in relation to the size of varices, portal venous pressure, portal systemic encephalopathy, and renal shunts.
Patients
and Methods
During the period from September 1980 to April 1984, 230 patients (175 men, 55 women, aged 29-79 yr) who were endoscopically found to have either esophageal varices, gastric varices, or both, were investigated by portal vein catheterization (4-6) in the First Department of Medicine, Chiba University Hospital and its affiliated hospitals. The hemodynamic study was done in these patients to assess whether a surgical procedure was indicated and, if so, which operation was preferred. The study group comprised 201 cases of liver cirrhosis, including 32 cases with hepatocellular carcinoma smaller than 5 cm with no intraportal invasion (studied by ultrasound), 20 cases of idiopathic portal hypertension (7,8), 4 cases of primary biliary cirrhosis, and 5 cases of miscellaneous diseases. The clinical status of the patients was Child’s A in 86 cases, Child’s B in 111 cases, and Child’s C in 33 cases not having ascites. Although 79 patients (72 men and 7 wom-
Abbreviations used in this paper: EV, esophageal varices; GR, gastrorenal; GV, gastric varices; PVP, portal venous pressure; SMV, superior mesenteric vein. 0 1988 by the American Gastroenterological Association 0016-5085/88/$3.50
HEMODYNAMICS IN GASTRIC VARICES
August 1988
Figure
1. Endoscopic appearance of GV. A. Mild varices of grade GV (+) seen in the cardia. B. Medium-sized running posteriorly from the cardia. C. Marked varices of grade GV (3+) looking like a submucosal
en) had a drinking habit, only 35 were assumed to have had alcoholic cirrhosis and the remainder most likely had posthepatitic cirrhosis as judged from the course of serum aminotransferase levels. Informed written consent was obtained from each patient before portal catheterization. The patients who had had an operation for portal hypertension were not included. There were 12 patients with chronic portal systemic encephalopathy characterized by episodic attacks of encephalopathy who had otherwise relatively well-maintained liver function tests (9-11). Endoscopy The endoscopic findings of esophageal varices were classified according to the General Rules for Recording Endoscopic Findings set by the Japanese Research Society for Portal Hypertension (12), in which the form of varices is graded as F, (straight), F, (winding), and F3 (nodule-beaded). The degree of gastric varices (GV) was classified into three grades according to Yasumoto (13): GV (+), mild varices in the cardia; GV (2+), winding veins seen in the cardia-fornix region; GV (3+), large varices in the cardia-fornix looking like a tumor (Figure 1). Percutaneous
Transhepatic
varices of grade GV (2+) tumor.
vein(s) that arise in the midportion of the splenic vein and run toward the cardia are called “posterior gastric,” and the veins that arise to the right of the midportion of the splenic vein and run toward the cardia are called “left or right gastric.” They used to be called the “coronary” veins. The collaterals that connect these gastric veins and the left renal vein studied on the venograms are called gastrorenal (GR) shunts. They were graded according to the diameter of the left suprarenal vein immediately before entering the left renal vein as follows: +, ~5 mm in diameter; 2+, 5-10 mm: 31, >lO mm (Figure 2). Similarly, splenorenal shunts were defined as the connections between the splenic vein and the left renal vein, and they were graded according to the diameter of the connecting vein shortly before it entered the left renal vein as follows f. x:5 mm; 2+, 5-10 mm: 3+, >lO mm. The flow direction of the contrast medium in these veins and collaterals was also studied. For the statistical analysis, either the ,$ test or the Student’s t-test (comparison of averages) was used.
Results Endoscopic
Catheterization
The ultrasound-guided procedure already detailed (6,14,15) was used. After measurement of portal venous pressure with the zero point at 10 cm above the table, contrast medium was injected through the catheter tip into the splenic hilum and into the superior mesenteric vein for the study of collateral veins and blood flow direction. By these venographic procedures, the routes of hepatofugal blood flow that formed varices were clarified. The sizes of collateral veins were measured on the films to determine the major route of portal-systemic shunting. In this study, the short gastric veins are defined as those veins that arise in the splenic hilum or within the spleen and course from the greater curvature of the stomach around the fornix. The
435
The five groups gastric
patients
were
depending
varices
assessment
Findings
(EV
divided
into the following
on the size of esophageal and
of the degree
GV,
respectively).
and In the
of EV and GV, F1 was taken
as equivalent to GV (+), F2 equivalent to GV (2+), and F, equivalent to GV (3+). Thus, the five groups were EV: EV > GV: EV + GV:
there were only EV (n = 99, 43%), EV were larger than GV (n = 98, 42.5%), EV were about the same in degree as GV (n = 10, 4.3%),
436
Figure
WATANABE
ET AL.
GASTROENTEROLOGY
2. Portographic features of three grades of GR shunt. A. Gastrorenal shunt (+). in which the diameter of the suprarenal vein entering the left renal vein is <5 mm. B. Gastrorenal shunt (2+), the suprarenal vein is between 5 and 10 mm. C. Gastrorenal shunt (3+), the suprarenal vein is >lO mm. C’. Late-phase film of C. GV. gastric varices; IVC, inferior vena cava: LRV, left renal vein; large arrow, the inferior diaphragmatic vein; two small orrows. the left suprarenal vein.
EV < GV: GV:
GV were larger than EV (n = 15, 6.5%), there were only GV (n = 8, 3.5%).
Thus, 131 of 230 patients degree. Venographic
(57%)
had GV of varying
Findings
The venographic findings in these 230 patients were divided into three groups: the right type (R > L), in which the left/right gastric vein was the predominant collateral route to form the varices; the left type (R < L), in which posterior/short gastric veins were the major route that formed the varices; and the equal type (R + L), in which the left/right R>L
R%L
gastric vein and the posterior/short gastric veins contributed about equally to the formation of varices (Figure 3). The majority, 179 cases, showed the right type (77.8%), 20 cases were of the equal type (8.7%), and 31 cases were of the left type (13.5%). As shown in Table 1, the majority of EV and EV > GV cases had the right type (83% and 86%, respectively]. By contrast, only 25% of GV cases and 33% of EV < GV cases had the right type, about one-half of GV and EV < GV cases showed the left type, and the remainder had the equal type.
Table
1. Endoscopic Grading of Esophageal and Gastric Varices in Relation to the Collateral Veins That Were Supplying Them
R
Figure
Vol. 95, No. 2
3. Three types of venous supply to gastric varices as classified by portography. R > L. right type; R + L, equal type; R < L, left type. EV, esophageal varices; GV, gastric varices; PV, portal vein; SV, splenic vein: L, posterior and/or short gastric vein; R. left or right gastric vein.
EV EV > GV EV + GV EV< GV GV Total
Types R>L 83 84 5 5 2
(83%) (86%) (50%) (33%) (25%)
179 (77.8%)
of venous
supply R
R+L 5 8 3 2 2
(5%) (8%) (30%) (13%) (25%)
20 (8.7%)
to varices
11 6 2 8 4
(11%) (6%) (20%) (53%) (50%)
31 (13.5%)
Total 99 98 10 15 8
(100%) (100%) (100%) (100%) (100%)
230 (100%)
EV. esophageal varices; GV, gastric varices; R. left or right vein: L, posterior or short gastric veins, or both.
gastric
HEMODYNAMICS
August
1988
Table
2. Relationship Between Grading of Esophageal and Gastric Varices and the Size of Gastrorenal Shunts Grade
Type
of varices
-
Frequency of GR shunt
+
2+
3+ 1
17 (17%) 21 (21%)
82
14
2
EV > GV (n = 981
77
14
4
3
4
1
1
4
6 (60%)
3
0
4
8
12 (80%)
0
0
1
7
8 (100%)
166
29
12
23
64 (27.8%
GV(n = 10) EV < GV (n = 15) GV (n = 6)
Total EV, esophageal
varices;
GV. gastric
Relationship Between Degree of Gastrorenal
varices;
Types Shunt
Relationship Between Types of Varices Portal Venous Pressure and Gastrorenal Shunts
and
and
PVP” (mm&O)
EV (n = 991 EV > GV (n = 981 EV + GV (n = lo] EV i GV (n = 15) GV (n = 8) EV, esophageal varices: GV, gastric pressure. ‘Mean 2 SD. ’p < 0.01.
326 t 66b 340 k 65 337 + 63 302 + 75 240 + 37b
varices;
PVP, portal
_
+
Yes
339 + 66”
343 ? 65
(n = 84)
(n f 15)
3+ 277 C 67”
131
2 -t 345 -t 56 ,
(n 7 10)
(11 = 22)
venous
77”.
(n = 4’7) No
3. Relationship Between Grading of Esophageal and Gastric Varices and Portal Venous Pressure of varices
GV
Total of cases
of GR shunt
311 t
As summarized in Table 3, the average portal venous pressure (PVP) was 326 + 66 mmHzO (mean ? SD] in the EV group, 340 t 65 mmHzO in the EV > GV group, 337 ‘- 63 mmHzO in the EV + GV group, 302 rt 75 mmHzO in the EV < GV group, and 240 +- 37 mmHsO in the GV group. The difference between the EV and GV groups was significant (p < 0.01). Among the cases with GV, the correlation between the grade of GR shunt and PVP was analyzed (Table 4). When 84 patient.s not having GR shunts and 47 having GR shunts were compared for PVP, the latter (311 -+ 73 mmHzO) had a significantly lower PVP compared with the former (339 +- 66 mmHzO) (p < 0.05). The portal venous pressure was
Type
437
Between the Degree of Shunt and Portal Venous Pressure
of total)
CR, gastrorenal
of Varices
VARICES
343 k 58b
Table 2 illustrates this relationship. Whereas only 17% of EV (EV only) cases and 21% of EV > GV cases had GR shunts, 80% of EV < GV cases and 100% of GV cases had GR shunts (EV and EV > GV combined vs. EV < GV and GV combined, p < 0.001). In these latter groups, the degree of GR shunt was also marked in the majority.
Table
4. Relationship Gastrorenal
Degree
of GR shunt
EV (n = 99)
EV+
Table
IN GASTKIC
326 + 66’.
324 + 59
(n = 82)
(n = 17)
GR, gastrorenal; 0.05 (differences bols).
99
GV, gastric varices. ” p < 0.05; ” p i 0.01: ’ p
in 22 patients who had GV and ‘_ 67 mmHzO (3s) GR shunts; it was significantly lower than that in any other group, even than that of the (+) and (2-t) groups combined (p < 0.01). Thus, it seemed that if GV became significantly large, PVP would decrease. No such differences were noted among groups without GV with regard to the size of GR shunts. Altogether there were 21 patients who had SR shunts in the whole series: 12 had (+) shunts, 5 (2+) shunts, and 4 (3+) shunts. The portal venous pressure in these groups was 328 + 67. 337 k 32, and 313 + 111 mmH,O, respectively. 277
Direction Varices
of Blood
Flow
in Relation
to
The blood flow from the superior mesenteric vein (SMV) into the portal or splenic vein, or both, could be divided into three types: A, most SMV blood flowed into the portal vein. opacifying the intrahepatic portal branches clearly: B, part of the SMV blood flowed into the splenic (and left gastric] vein with a resultant reduced flow into the portal vein and poor opacification of intrahepatic portal branches; and C, all SMV blood flowed into the splenic and left gastric veins without opacification of the portal vein and its tributaries, as illustrated in Figure 4. Table 5 correlates the five types of varices and the three blood flow types. Type A flow was the predominant type in patients with EV only (EV) or mainly esophageal varices (EV > GV). As GVs became large (groups EV + GV, EV < GV. and GV), there were more type B flows (40%, 33% and 63%, respectively); 2 of 8 (25%) cases in the GV group had total diversion of SMV flow away from the liver (type C). In 7 of 8 cases (88%), the flow in the splenic vein near its junction with the SMV was retrograde.
438
WATANABE
ET AL.
GASTROENTEROLOGY
Type A
Type B
I
Table
Type C
I
6. Frequency in Relation
of Portal Systemic Encephalopathy to Grading of Esophageal and Gastric Varices
Direction of portal venous Type of varices I
Visualization of IHPB Figure
good
poor
“0
4. Schematic representation of the three types of portal hemodynamics with respect to the flow of superior mesenteric venous blood. IHPB, intrahepatic portal branches.
Relationship of Varices and Chronic Portal Systemic Encephalopathy Of the 230 patients studied, 12 had a typical (9-11). Five chronic portal-systemic encephalopathy of the 12 encephalopathic patients had marked GV (41.7%). Therefore, the relationship between varices and encephalopathy was investigated. As shown in Table 6, chronic encephalopathy was uncommon among patients of the EV and EV > GV groups [3.0% and 4.l%, respectively], whereas patients in the EV < GV and GV groups had portal systemic encephalopathy more frequently (12.0% and 25.0%, respectively). The difference between the EV and GV groups was significant (p < 0.011, and similarly there was a significant difference between the EV and EV > GV groups combined and the GV and EV < GV groups combined (p < 0.05). Encephalopathy was also correlated with the type of SMV blood flow. It occurred in 66.7% of the type C group, 30.8% of the type B group, and 1% of the type A group [Table 7).
Discussion In the barium swallow examination of the stomach, demonstration of GV needs certain uncom-
Table
5. Relationship
Between Grading of Esophageal and Gastric Varices and Types of Portal Hemodynamic Changes Shown in Figure 4 Type of hemodynamic
Type of varices EV EV > GV EV + GV EV < GV GV
B
A 95 89 6 10 1
(96%) (90%) (60%) (67%) (12%)
changes”
3 9 4 5 5
(3%) (10%) (40%) (33%) (63%)
1 0 0 0 2
C
Total
(1%) (0%) (0%) (0%) (25%)
99 98 10 15 8
EV, esophageal varices; GV, gastric varices. ’ A, good opacification of intrahepatic portal branches: B, poor opacification of intrahepatic portal branches; C, no opacification of intrahepatic portal branches (total hepatofugal flow of superior mesenteric venous blood).
Vol. 95, No. 2
EV EV EV GV
(n = 99) > GV (n = 98) 5 GV (n = 25) (n = 8)
Frequency of portal systemic encephalopathy 3 (3.0%)” 4 (4.1%) 3 (12.0%) 2 (25.0%)”
EV, esophageal varices; GV, gastric varices. ’ p < 0.01.
mon projections, without which they are often missed (13,161. In this respect, endoscopy is more diagnostic than the routine barium examination. With a second barium examination specifically aimed at detection of GV, Yasumoto (13) demonstrated GV in 55.2% of patients with cirrhosis, 63.2% of patients with cirrhosis and hepatocellular carcinoma, and 90.9% of patients with idiopathic portal hypertension. In the present study, the incidence of GV seen by the endoscope was 57%, which is close to the figures obtained by Yasumoto with the second barium study. However, in a study by one of the current authors (17), endoscopy failed to demonstrate GV in 9 of 46 patients with roentgenologically demonstrable GV. Perhaps neither procedure is perfect in the detection of small GV that may not be clearly distinguishable from normal rugae by endoscopy. The hemodynamic study of these patients with GV clearly showed that GV are frequently supplied from the short and posterior gastric veins in contrast to EV, which are known to be supplied mainly from the coronary (left and right gastric) vein. As the connection between GV and the portal vein through the short and posterior gastric veins is short, therapeutic sclerotherapy carries a risk of developing portal thrombosis. In patients with large GR shunts, a sclerosing agent could even go into the renal shunt, Table
7. Frequency of Portal Systemic Encephalopathy in Relation to the Type of Portal Hemodynamic Changes Type of portal hemodvnamic changes”
Frequency of portal systemic enceuhalouathv
A (n = 201) B (n = 26) C (n = 3)
2 (1.0%) 8 (30.8%) 2 (66.7%)
aA, good
opacification of intrahepatic portal branches; B, poor opacification of intrahepatic portal branches; C, no opacification of intrahepatic portal branches due to total diversion of superior mesenteric venous blood away from the liver (see Figure 4).
August
1988
and from there into the inferior vena cava. In view of the lower incidence of rupture, no prophylactic sclerosis seems warranted for gastric GV. They are covered by a relatively thick gastric mucosa as compared with EV, and rupture is uncommon unless a peptic ulcer develops immediately above them. What causes the major portal-systemic collateral route to develop as PVP increases? It could be purely anatomic, depending on the individual venous vascular makeup, or be due to a preexisting anomalous vein. The size of the potential collateral vein and the length of the connection by that vein between the portal axis vein and the systemic (caval) veins will also be a factor. At the moment, there is no way of predicting the major route of shunting in the early stage of portal hypertension, hence artificially altering the route to prevent future variceal bleeding. The so-called red color sign (12), which portends future rupture and which is frequently seen over the surface of EV, was not observed in GV. Whereas the former involve veins in the lamina propria and mucosa (18-20).the same does not seem to occur in the latter. Perhaps due to the vascular anatomy in the cardia of the stomach, the short and posterior gastric veins connect more often with the left renal vein than does the left gastric vein, to form large shunts through the inferior diaphragmatic and the suprarenal vein. The portal venous pressure was significantly lower in patients with large GV and in patients with large GR shunts. There are occasional reports on normal portal pressure in the presence of a huge shunt (21) and reduction of PVP with enlargement of the renal shunt (22).Ohnishi et al. (23)studied the relationship between PVP and the degree of shunts in 52 patients with portal hypertension and found a correlation of a quadratic linear regression. In other words, PVP tended to decline as shunting became more extensive. The reason for the lower PVP observed in our patients with large GV and large GR shunts may be the same. Failure to demonstrate the same trend in patients with large splenorenal shunts was perhaps due to the small number of patients. Chronic portal systemic encephalopathy is a clinical syndrome in which the patient not severely debilitated by chronic liver disease develops encephalopathy in episodic bouts (g-11). It is commonly associated with large shunts, more often through the left renal vein than through other collateral routes (11,24-26).This was also observed in the current study with 12 patients. In such patients, the whole or part of the blood going out of the SMV flowed retrogradely into the splenic vein. Nitrogenous substances absorbed from the gut into the SMV enter the systemic circulation without going through, hence not being absorbed by, the liver in
HEMODYNAMICS
IN GASTRIC
VARICES
439
such patients. Thus, this syndrome should be suspected in patients with large GV or large renal shunts that can be seen by current real-time ultrasonograph (27).In this series, only 3% of the patients who had only EV had chronic portal-systemic encephalopathy. These data indicate that (a) GV occur nearly as frequently as EV but large GV are rather infrequent, (b) GV are different from EV in the route of blood supply, and (c) large GV are often associated with large GR shunts and chronic portal-systemic encephalopathy.
References M, Feldman M Jr. Gastric varices. Gastroenterology 1. Feldman 1956;30:318-21. 1953;60:46-50. 2. Evans JA, Delany F. Gastric varices. Radiology importance of gastric varices. N 3. Karr S. Wohl GT. Clinical Engl J Med 1960:263:665-9. A. Vang J. Transhepatic catheterization and 4. Lunderquist obliteration of the coronary vein in patients with portal hypertension and esophageal varices. N Engl J Med 1974;291: 646-9. K. Suzuki K. Musha H. Arimizu N. Percutaneous 5. Okuda transhepatic catheterization of the portal vein for the study of portal hemodynamics and shunts: a preliminary report. Gastroenterology 1977:73:279-84. K. Percutaneous transhepatic fi.Okuda K, Kimura K, Takayasu portography and sclerotherapy. Semin Liver Dis 1982;Z: 5 7-66. K, Omata M, eds. Idiopathic portal hypertension. 7. Okuda Tokyo: University of Tokyo Press, 1983. 8. Okuda K. Kono K, Onhishi K, et al. Clinical study of 86 cases of idiopathic portal hypertension and comparison with cirrhosis with splenomegaly. Gastroenterology 1984;86:600-10. 9. Sherlock S, Summerskill WHJ, White LP. Portal-systemic encephalopathy-neurological complications of liver disease. Lancet 1954;ii:453-7. 10. McDermott WV Jr, Adams R. Episodic stupor associated with an Eck fistula in the human with particular reference to the metabolism of ammonia. J Clin Invest 1954;33:1-9. 11. Summerskill WHJ, Davidson EA. Sherlock S. Steiner RF. The neuropsychiatric syndrome associated with hepatic cirrhosis and an extensive collateral circulation. Q J Med 1956;25: 245-66. 12. Japanese Society for Portal Hypertension. The general rules for recording endoscopic findings on esophageal varices. Jpn J Surg 1980;10:84-7. 13. Yasumoto M. Clinical observations on 100 cases of gastric varices. Jpn J Gastroenterol 1971:68:721-39. 14. Kimura K. Ohto M, Tsuchiya Y. et al. Single puncture method for percutaneous transhepatic portography using a thin needle. Radiology 1981;139:748-9. 15. Ohto M, Karasawa H, Tsuchiya Y. et al. [Jltrasonically guided percutaneous contrast medium injection and aspiration using a real-time puncture transducer. Radiology 1980:136:171-6. 16. Marshall JP II, Smith PD. Hoyumpa AM Jr. Gastric varices. Problem in diagnosis. Dig Dis 1977:22:947-55. 17. Okuda K, Yasumoto M, Goto A. Kunisaki T. Endoscopic observations of gastric varices. Am 1 Gastroenterol 1973;60: 357-65. 18. Kegaries DL. The venous plexus of the esophagus. Surg Gynecol Obstet 1934;58:46-51.
440
WATANABE
ET AL.
19. Butler H. The veins of the esophagus. Thorax 1951;6:276-96. 20. Noda T. Angioarchitectural study of esophageal varices. Virchows Arch [A] 1984;404:381-92. 21. Takashi M, Igarashi M, Hino S, Goto N, Okuda K. Chronic portal-systemic encephalopathy with normal portal vein pressure possibly due to noncirrhotic portal fibrosis. Dig Dis Sci 1984;29:669-73. 22. Ohnishi K, Nakayama T, Nomura F, Koen H, Okuda K. Reduction of portal vein pressure with the enlargement of portal-systemic shunts: observations made in one patient. J Clin Gastroenterol 1984;6:447-51. 23. Ohnishi K, Nakayama T, Koen H, et al. Interrelationship between type of spontaneous portal systemic shunt and portal vein pressure in patients with liver disease. Am J Gastroenterol 1985;80:5614. 24. Takashi M, Igarashi M, Hino S, et al. Portal hemodynamics in chronic portal-systemic encephalopathy. Angiographic study in seven cases. J Hepatol 1985;1:467-76.
GASTROENTEROLOGY
Vol. 95, No. 2
25.
Ohnishi K, Sato S, Saito M, et al. Clinical and portal hemodynamic features in cirrhotic patients having a large spontaneous splenorenal and/or gastrorenal shunt. Am J Gastroenterof 1986;81:450-5. 26. Lam KC, Juttner HU, Reynolds TB. Spontaneous portosystemic shunt-relationship to spontaneous encephalopathy and gastrointestinal hemorrhage. Dig Dis Sci 1981:26:34652. 27. Takayasu K, Moriyama N, Shima Y. et al. Sonographic detection of large spontaneous spleno-renal shunts and its clinical significance. Br J Radio1 1984;57:565-70.
Received July 13, 1987. Accepted March 9, 1988. Address requests for reprints to: Kunio Okuda, M.D., Ph.D., First Department of Medicine, Chiba University Hospital, Chiba, Japan (280).