0016-5107/90/3602-0S21$02.00 GASTROINTESTINAL ENDOSCOPY Copyright © 1990 by the American Society for Gastrointestinal Endoscopy
Assessment of portal hypertension by endoscopic ultrasonography Giancarlo Caletti, MD, Emilio Brocchi, MD Mario Baraldini, MD, Alberto Ferrari, MD Maria Gibilaro, MD, Luigi Barbara, MD Bologna, Italy
Endoscopic ultrasonography (EUS) was performed in 40 patients with portal hypertension (PH) and in 48 control subjects. The azygous, splenic, mesenteric, and portal veins were displayed in both groups. However, esophageal and gastric varices, periesophageal and perigastric collateral veins, and submucosal gastric venules were displayed only in patients with portal hypertension. EUS was inferior to endoscopy in detecting and grading esophageal varices (p < 0.0005), but EUS was superior in the detection of varices in the fundus of the stomach (p < 0.0005). Detection of periesophageal veins by EUS increased with increasing diameter of esophageal varices at endoscopy (57% in grade 1,89% in grade 2, and 100% in grade 3), and there was a direct correlation between endoscopic grade and the diameter of the periesophageal collateral veins at EUS. The diameter of the azygous vein by EUS at its distal and proximal margins was significantly greater in patients with PH (p < 0.001); the EUS diameter of the azygous vein was significantly larger with variceal grade 2 compared with grade 1 (p < 0.02 and p < 0.01, respectively). In portal hypertensive gastropathy, endoscopic and EUS detection were coincident. No correlation was found between the presence of portal hypertensive gastropathy, endoscopic grade of esophageal varices, and detection of gastric varices at EUS. (Gastrointest Endosc 1990;36:S21-S27)
We have demonstrated that endoscopic ultrasonography (EUS) can display esophageal and gastric varices and also periesophageal collateral veins. 1• 2 More recently, we have shown that EUS can demonstrate small dilated vessels within the gastric wall in patients with portal hypertensive gastropathy.~ Since other vascular structures of the portal venous system are also displayed by EUS, we undertook a prospective study to determine which structures of the portal system are visualized by this technique in patients with portal hypertension (PH) and in normal subjects. This study also attempts to correlate endoscopic and EUS findings in order to determine whether EUS contributes to the assessment of PH.
Received December 1, 1989_ Accepted JanUllry 23, 1990. From the Academic Department of Medicine and Gastroenterology, University of Bologna, Policlinico S. Orsola, Bologna, Italy. Reprint requests: Giancarlo Caletti, MD, Viale Gozzadini 11/2, 40124 Bologna, Italy. VOLUME 36, NO. 2, 1990
MATERIALS AND METHODS
Upper gastrointestinal endoscopy and EUS were obtained in 40 consecutive patients with PH due to liver cirrhosis and in 48 control subjects between June 1988 and June 1989. The presence of PH and esophageal varices were confirmed by abdominal ultrasonography and endoscopy in all cirrhotic patients. This group, consisting of 26 men and 14 women, had a mean age of 56 years (range, 28 to 74 years). The etiology of the cirrhosis was post-necrotic in 12 patients, alcohol in 8, and unknown in 20. The Child-Pugh classification of the severity of liver disease was A in 10, B in 28, and C in 2 patients. In the control group, the mean age was 55 years (range, 31 to 70 years) and there were 28 men and 20 women. Upper gastrointestinal endoscopy and EUS were obtained in control subjects because of various benign gastric conditions including submucosal lesions and peptic ulcer. There was no clinical, laboratory, or ultrasonographic (transabdominal) evidence of liver disease or PH in the control group, and there was no endoscopic evidence of esophageal or gastric varices. EUS was performed with the Olympus GF-UM3/EU-M3 821
ultrasound fiberscope and display unit. This system provides a 360-degree radial scan with frequencies of 7.5 MHz or 12 MHz. Both frequencies were always utilized in the examination of the various structures. Details of our EUS technique have been described previously"'" EUS examination of the esophagus and surrounding structures is performed with a water-filled balloon over the ultrasonic probe; EUS exploration of the stomach and adjacent structures is conducted by instillation of deaerated water into the stomach. After the distal end of the ultrasound fiberscope is placed at the pylorus, the EUS examination is begun by slowly withdrawing the insertion tube of the instrument. It is well established that the echographic structure of the esophageal and gastric walls, which always consists of five layers of different echogenicity, corresponds closely to the actual anatomical architecture. 6 . 7 The first (hyperechoic) layer corresponds to the fluid/mucosal interface. The second (hypoechoic) layer is probably produced in the deep portion of the mucosa and includes the muscularis mucosae (which is too thin to generate a separate interface). The third (hyperechoic) layer corresponds to the submucosa and the submucosa/muscularis propria interface. The fourth (hypoechoic) layer is created as the ultrasound beam penetrates the muscularis propria. The fifth (hyperechoic) layer corresponds mainly to the serosa and periserosal fat. The length and the diameter of the visualized vascular structures of the portal venous system were measured and recorded. The results are given as means ± 1 SD. Correlation of endoscopic and EUS findings and between different EUS findings were performed using non-parametric procedures.
RESULTS Endoscopy in patients with PH
Esophageal varices were graded from 1 to 3 according to their size in the 40 patients with PH. 8 Grade I varices were seen in 14 patients (35%), grade 2 in 18 (45%), and grade 3 in 8 (20%). The distance from the incisor teeth to the distal margin of the varices ranged from 35 to 44 cm (mean 39.1 ± 2.4); the distance to the proximal margin ranged from 16 to 37 cm (mean 25.5 ± 6.5). The range in length of the variceal columns was 4 to 20 cm (mean 13.8 ± 5.8). Gastric varices, scored as present or absent, were found at endoscopy in 10 patients (25%) including 4 patients with grade 2 esophageal varices (22.2%) and 6 patients with grade 3 varices (75%). Gastric varices were not seen in any ofthe 14 patients with grade 1 esophageal varices. Endoscopic findings of portal hypertensive gastropathy have been described. 9 , 10 These changes were found in eight patients (20%). Three of them had grade 1 esophageal varices, two had grade 2, three had grade 3, and four had gastric varices.
(14%) with grade 1 esophageal varices, 14 of 18 with grade 2 (78%), and 4 of 8 with endoscopic grade 3 (50%) (Fig. 1). The distance from the incisor teeth to the distal margin of the varices ranged from 35 to 44 cm (mean 39.1 ± 2.3), while that to the proximal margin ranged from 26 to 37 cm (mean 32.4 ± 4). The length of the variceal columns ranged from 2 to 14 cm (mean 6.7 ± 4.3). The diameter of the varices at the level of the cardia was 2 to 4 mm (mean, 2.6 ± 0.6); at the proximal margin, the range in diameter was 2 to 4 mm (mean 2.6 ± 0.6). Periesophageal collateral veins, which appeared in transverse section as rounded, echo-free structures just external to the esophageal wall, were noted in 32 patients (80%). These external collateral vessels were demonstrated in 8 patients (57%) with grade 1 varices, 16 with grade 2 (89%), and 8 of 8 patients with grade 3 esophageal varices (Fig. 2). The distance from the incisor teeth to the distal margin of the collateral veins ranged from 35 to 44 em (mean, 39.4 ± 2.5); the distance to the proximal margin was between 20 and 42 em (mean, 32.2 ± 6.9). The overall length of the collateral veins ranged from 1 to 16 em (mean 7.2 ± 5.5). These veins measured between 2 and 7 mm in diameter (mean, 4.6 ± 1.6) at the level of the cardia; the diameter at the proximal margin was 2 to 7 mm (mean, 4.5 ± 1.6). The azygous vein, seen in all 40 patients, appeared in transverse section as a rounded, echo-free structure between the esophageal wall, aorta, and spine (Fig. 3). The proximal margin is established when the vein arches in a forward and slightly lateral direction at the level of the fourth thoracic vertebra to enter the superior vena cava. The distance from the incisor teeth to the distal margin of the vein ranged from 28 to 38 em (mean, 31.8 ± 2.7), while that to the proximal
EUS in patients with PH
Esophageal varices were displayed in transverse section as rounded, echo-free structures just beneath the mucosal and submucosal layers in only 20 of the 40 patients. EUS demonstrated varices in 2 of 14 patients 822
Figure 1. Transverse 360-degree EUS scan of the esophagus at the junction with the cardia. Large submucosal varices (V) are clearly displayed. A, aorta; E, esophageal wall.
GASTROINTESTINAL ENDOSCOPY
scopic grade 1 esophageal varices, 10 (55.5%) with grade 2, and 8 (100%) with grade 3. EUS displayed numerous small, rounded echo-free structures within the submucosa of the stomach in all eight patients with endoscopic signs of portal hypertensive gastropathy (Fig. 5). This finding was not present in the other 32 cirrhotic patients in whom the stomach was normal at endoscopy. Dilated vessels were demonstrated as numerous echo-free structures just external to the gastric wall in 18 patients (45 %). The diameter of these perigastric veins was 2 to 3 mm (mean, 2.3 ± 0.2). The splenic vein was displayed as a longitudinal echo-free channel posterior to the stomach and pancreas in all patients. Its complete course from the Figure 2. Transverse 360-degree EUS scan of the esophagus
at the junction with the cardia. Dilated periesophageal veins (V) are visualized outside the wall. A, aorta.
Figure 4. EUS scan of the fundus using the water-filled stomach technique. Gastric varices (V) are clearly visualized. S, stomach.
Figure 3. EUS 360-degree scan of the esophagus. The azygous vein (Z) is visualized with the esophageal wall (E), the aorta (A), and the spine (sp).
margin ranged from 23 to 30 cm (mean, 25.9 ± 2.2). The length of the azygous vein was from 3 to 13 cm (mean, 5.9 ± 2.4). The diameter at the distal margin ranged from 5 to 13 mm (mean, 7.9 ± 1.8); the range at the proximal margin was 5 to 14 mm (mean, 9.2 ± 2.1). With the ultrasound probe positioned below the gastroesophageal junction and the stomach filled with water, EUS displayed rounded, echo-free structures beneath the mucosa and submucosa in all 10 cases in which gastric varices were seen at endoscopy (Fig. 4). This same echo pattern of gastric varices was also found in 12 patients with no endoscopic evidence of varices in the stomach. Overall, EUS detected gastric varices in 22 patients (55%); 4 (28.5%) with endoVOLUME 36, NO. 2, 1990
Figure 5. A360-degree EUS scan of water-filled stomach (S).
Dilated veins within the gastric wall (arrows) are seen. In this patient, endoscopic findings of portal hypertensive gastropathy were present. 823
spleen to the porta was seen in all 40 cases. Its maximum diameter was measured at 12 to 16 mm (mean, 13 ± 0.1). The mesenteric vein was noted in all patients as a round, echo-free structure posterior to the gastric wall and pancreas. Its maximum diameter ranged from 11 to 15 mm (mean, 12 ± 0.2). With the ultrasonic probe located in the upper part of the body, the confluence of the splenic and the superior mesenteric veins to form the portal vein was noted external to the lesser curvature of the gastric wall in 37 patients (92.5%). The portal vein was partially visualized in only 11 patients (27.5%), and its entire course was never visualized. Its maximum diameter measured at the confluence of the splenic and the superior mesenteric veins ranged from 14 to 18 mm (mean, 15 ± 0.1). EUS in control patients
EUS did not display any of the findings or structures referable to esophageal varices, periesophageal collateral veins, gastric varices, or portal hypertensive gastropathy in any of the 48 control subjects. The azygous vein was noted in all normal subjects. The distance from the incisor teeth to the distal margin of the vein ranged from 27 to 36 cm (mean, 31.3 ± 2.3); the distance to the proximal margin was between 23 and 30 cm (mean, 26.2 ± 1.8). The length of the portion of the azygous vein visualized was from 2 to 9 cm (mean, 5.2 ± 1.8). The diameter at the distal margin was 4 to 9 mm (mean, 5.9 ± 1.2); the diameter at the proximal margin was between 5 and 9 mm (mean, 6.8 ± 1.1). A single, rounded echo-free structure external to the greater curvature of the gastric body was noted in 19 patients (39.6%). The diameter of this solitary perigastric vessel was 2 to 3 mm (mean, 2.1 ± 0.4). The splenic vein was demonstrated in all control subjects. It ranged in diameter from 7 to 11 mm (mean, 9 ± 1.3). The mesenteric vein was noted in 43 control subjects (89.5%). It had a maximum diameter of from 6 to 10 mm (mean, 8 ± 1.5). The origin of the portal vein at the confluence of the splenic and the superior mesenteric veins was demonstrated in 43 individuals (89.5%). The entire course of the portal vein was never visualized, although it was partially displayed in 13 subjects (27%). The range of maximum diameter was 9 to 12 mm (mean, 11 ± 1.2).
The detection of periesophageal veins by EUS increased with increasing variceal diameter at endoscopy (57% in grade 1, 89% in grade 2, and 100% in grade 3) (Fig. 6), and there was a direct correlation between endoscopic grade and the diameter of the periesophageal veins at EUS (p < 0.025, Spearman's rank correlation test). A direct correlation was also found between the length of varices at endoscopy and length of periesophageal veins at EUS (p < 0.001, Spearman's rank correlation test). EUS was superior to endoscopy in the detection of gastric varices (p < 0.0005, Fisher's exact test). The rate of detection increased in relation to the endoscopic grade of esophageal varices. When the diameter of the periesophageal veins as measured by EUS was compared in cirrhotic patients with and without gastric varices, no difference was found. There was also no difference found when the EUS diameter of the azygous vein at its distal and proximal margins was compared in cirrhotic patients with and without gastric varices. There was no difference in the ability of EUS and endoscopy to detect portal hypertensive gastropathy. There was no correlation between the presence of portal hypertensive gastropathy, endoscopic size of esophageal varices, and the presence of gastric varices at EUS. There was also no correlation found between the presence of portal hypertensive gastropathy and the EUS diameter of the distal and proximal margin of the azygous vein. There was also no correlation between normal subjects and patients in terms of the detection of perigastric veins by EUS. The distances from the incisor teeth to the distal and proximal margins of the azygous vein did not differ significantly in cirrhotic patients and control 100
100
75
7. 50
25
Relationships
Endoscopic and EUS findings were related with respect to the size and length of esophageal varices. Endoscopy was not only superior to EUS in detecting esophageal varices (p < 0.0005), but it was also more accurate with respect to grading (p < 0.0005 for grade 1, not significant for grade 2; p < 0.05 for grade 3) and in assessing their length (p < 0.001). 824
o Grode 1
Grode 2
Grode 3
Endoscopic grading of esophageal varices
Figure 6. EUS in the detection of periesophageal veins. The ability to visualize these structures increases significantly with increasing diameter of esophageal varices. GASTROINTESTINAL ENDOSCOPY
subjects. However, the diameters at the distal and proximal margins of the azygous were significantly greater in the cirrhotic group (p < 0.001, Wilcoxon rank sum test for independent data) (Fig. 7). There was no correlation between the diameter of the azygous vein (distal and proximal margins) and the diameter of the periesophageal veins. The diameter of the azygous vein at its distal and proximal margins was significantly greater in grade 2 esophageal varices compared with grade 1 (p < 0.02 and p < 0.01, respectively, Wilcoxon rank sum test for unpaired data) (Fig. 8); there was no significant difference when patients with grade 3 varices were compared with those with grade 2. DISCUSSION
The portal venous system can be visualized in a number of ways in patients with portal hypertension. Angiography has been the leading technique, although less invasive methods have been proposed including dynamic CT scanning. l l ,12 Abdominal ultrasonography is a suitable screening method for portal hypertension;13,14 measurements of the main portal vessels are reliable and a diagnosis of portal obstruction can be made in a high percentage of patients. 1:3- 18 Unfortunately, small tributaries are seldom visualized and no information is provided about the gastroesophageal collateral veins which anastomose with the caval system. 19 Recently, color-flow Doppler sonography has emerged as a promising technique for flow studies.20,21 Nuclear magnetic resonance scanning is practical for the morphologic assessment of the portal venous system. 22 EUS visualizes a large part of the portal venous system and it is a safe, well-tolerated, relatively non-
invasive technique that may be repeated numerous times. 4,5 The azygous, splenic, mesenteric, and portal veins are displayed in normal subjects and patients with PH. However, esophageal and gastric varices, periesophageal and perigastric collateral veins, and submucosal gastric venules are seen only with PH as these structures are normally very small and undetectable by EUS in normal subjects. We believe that EUS does not supplant transabdominal ultrasonography in the study of the splenic, mesenteric, and portal veins because the latter technique is less time consuming and provides better resolution. In fact, it is not possible to scan the entire course of the portal system by EUS. However, EUS detected fundal varices with greater frequency than endoscopy. There was a direct correlation between variceal grade at endoscopy and the detection by EUS of gastric varices. Endoscopy remains the most important technique for the assessment of esophageal varices. It is superior to EUS for detection of varices and in grading of these vessels according to diameter and length. This may be in part attributable to problems with focusing the ultrasound display and to compression of the varices by the water-filled balloon on the ultrasound fiberscope. Furthermore, EUS does not demonstrate the so-called red color signs that indicate an increased risk of bleeding.8, 2:3 In our experience, however, EUS has some value in confirming the eradication of esophageal varices and in assessing the patency and diameter of periesophageal collateral veins after a course of sclerotherapy.2,24 There was a correlation between the presence of periesophageal collateral veins and their diameter at EUS with the endoscopic grade of esophageal varices. These two systems are known to be connected by
LOWER LEVEL
UPPER LEVEL
p