Endoscopic approach to patients with portal hypertension: a complex diagnosis

Surgical Endoscopy

Surg Endosc (1990) 4:76-79

© Springer-Verlag 1990

Endoscopic approach to patients with portal hypertension: a complex diagnosis A retrospective study based on 10 years' experience R. V. Buccino, G. Bogliolo, M. Ferrara, V. Pietropaolo, L. Pecchioli, G. Miscusi, and A. Montori III CattedraPatologiaSpecialeChirurgica,PoliclinicoUmbertoI, Universit5degli Studi di Roma "La Sapienza",1-00161 Rome,Italy

Summary. We analyzed the endoscopic findings in 788 patients with esophageal and gastric varices who underwent upper gastrointestinal endoscopy between 1 January 1979 and 31 December 1988. Of these, 154 patients (19.6%) had gastric varices associated in various patterns with esophageal varices. Congestive gastropathy, occurring with esophageal and gastric varices (43.4%), was the most frequent pathology detected in our patients. Esophagitis was present in 15.8% of patients, but did not correlate with variceal bleeding. Endoscopy performed at 1 day to 1 week post-hemorrhage in 313 patients accurately identified the source of bleeding in only 57.2% of patients. This figure increased to 98.2% when we performed the examination within the first 24 h of hemorrhage. In this group varices were the source of hemorrhage in 72.3% of patients while the hemorrhage came from other sources, such as erosive gastritis, duodenal and gastric ulcer in 27.6% of patients. Key words: Endoscopy - Portal hypertension Esophageal varices - Gastrointestinal hemorrhage

In patients with portal hypertension, detection of esophageal varices is imperative, as they represent the most serious complication of this condition. Upper gastrointestinal endoscopy is the most widely used procedure for diagnosis of esophageal varices and succeeds in distinguishing variceal bleeding from other sources of bleeding in almost all patients. Furthermore, this procedure is the most appriopriate for investigating the range of gastrointestinal pathology that may accompany esophageal varices in patients with portal hypertension and may be responsible for bleeding and influence the therapeutic choices. The present study reviews our experience with endoscopy in patients with esophageal and gastric varices over a 10-year period.

Offprint requests to:

M. Ferrara, Via Flaminia,76, 1-00196 Rome, Italy

Patients and methods The records of 788 patients with esophageal and gastric varices examined endoscopicallyin our Digestive Endoscopy Center between 1 January 1979 and 31 December 1988 were reviewed. The mean age of our patients was 59 years (range 18-83); 67.2% were male and 32.8% female. Alcoholwas the predominantetiologyof portal hypertensionin this group of patients. Esophageal varices were classified into three groups: small, medium and large. Three hundred and thirteen patients underwent endoscopywithin a week of bleeding. Among these, 112 patients were examinedendoscopicallywithinthe first 24 h of bleeding. We gave 5 - 10 mg of Valiurn and 20 mg of Buscopani. v. as a premedication for endoscopic examination.The OlympusGIF K3, IT and Q 10 esophagogastrofiberscopeswereused.

Results Among the 788 patients with varices, 634 (80.4%) had esophageal varices alone, 137 (17.4%) had esophageal and gastric varices and 17 (2.2%) had only gastric varices. Esophageal varices were classified as: small, 185 (24%); medium, 251 (32.6%); large 335 (43.4%). Other gastrointestinal pathology was found in 547 patients (69.4%). In 342 patients (43.4%) with esophageal and gastric varices, congestive gastropathy was detected, particularly in patients with large varices. Of 125 patients (15.8%) with esophagitis, 67 had no history of previous hemorrhage. Duodenal ulceration was found in 89 patients (11.2%) while gastric ulcer and erosive gastritis occurred in 54 patients (6.8%) and 36 patients (4.5%), respectively. Among the 313 patients who had bled previously, 105 (33.5%) presented with other potential gastrointestinal causes of bleeding (e. g., erosive gastritis, gastric ulcer and duodenal ulcer). In 201 out of 313 patients, endoscopic examination was performed 24 h to 1 week after the hemorrhage. In this group the source of bleeding was identified without doubt in 115 patients (57.2%). The remaining 112 patients were examined endoscopically within the first 24 h of hemorrhage. In this group the bleeding source was found in 98.2% of the patients and identified it as being from esoph-

Table 1. Endoscopic findings in 112 patients with upper gastrointestinal hemorrhage who underwent early examination Variceal bleeding Visible bleeding Hemorrhage stigmata Diagnosis by exclusion Other sources Erosive gastritis Duodenal ulcer Gastric ulcer

ageal varices in 79 patients (70.5%). Of these, 25 patients (22.3%) had actively bleeding varices at the time of examination, while in 54 patients (48.2%) we observed the stigmata of bleeding (clots or fibrinoid aggregate) from varices. In only 2 cases (1.8%) did we assume the varices to be the source of hemorrhage as we found no other lesions (Table 1). In this group with bleeding esophageal varices, small varices were present in 2 patients (2.5%), medium in 27 patients (33.3%), and large in 52 patients (64.2%). Of the remaining 3 1 patients who underwent endoscopy within 24 h the source of hemorrhage was erosive gastritis (16 cases 14.3%), duodenal ulcer (10 cases 8.9%) and gastric ulcer (5 cases 4.5%) (Fig. 1).

Discussion In patients with esophageal varices, a careful search for gastric varices is indicated, as they may be a source of hemorrhage and result in a different therapeutic approach being adopted [18]. In our series 19.6% presented with gastric varices and in the vast majority of cases (89.9%) they were found in association with esophageal varices of larger caliber. Furthermore, in most cases the gastric varices were continuous with the esophageal varices at the cardia. The patients who presented with gastric fundal varices alone constituted a small minority.

Gastric ulcers

~denalulcers

Gastric erosions

Fig. 1. Sources of hemorrhage in 112 patients with varices who underwent endoscopic examination in the first 24 h after admission

It is difficult to comment on these results because of the small number of published reports dealing with the incidence of gastric varices detected endoscopically. Watanabe et al. [20] reported an incidence of gastric varices of 57%, while Hosking et al. [6] detected gastric varices in only 6% of their patients. This discrepancy can be explained by differences in patient selection and difficulty with endoscopic diagnosis. Okuda et al. [9] in a series of 46 patients with gastric varices detected radiologically, reported a 20% false-negative rate for endoscopic examination. Small isolated gastric varices are difficult to diagnose endoscopically and can be distinguished only with difficulty from normal mucosal folds of gastric fundus. This is especially the case in patients with segmental portal hypertension (e. g., splenic vein thrombosis) where the collateral circulation through the short gastric veins involves the gastric fundus and body, but not the esophagus. The fact that the esophagus is not involved can be explained by the absence or insufficiency of venous anastomoses between the gastric subglandular venous plexus and the intrinsic esophageal venous circulation. Vianna and coworkers [19] demonstrated the presence of small veins arranged in a "palisade" situated between the vein of the gastric fundus (tributary of portal circle) and the esophageal submucosal veins (tributary of the azygos vein) at the level of the esophagogastric junction. This "palisade" might act as a high resistance zone to the venous flow, preventing the dissemination of increased venous pressure from the subglandular gastric veins to the submucosal esophageal veins and vice versa. Therefore, increase in venous pressure caudal to this "palisade" zone would cause gastric variceal formation, while increased pressure above the "palisade" zone would cause the formation of submucosal esophageal varices by means of the incompetent perforating esophageal veins [7]. The presence of this palisade zone acting as a resistence to the venous flow between esophagus and stomach may support the observation made by us and others [6] that sclerotherapy of esophageal varices does not cause or worsen gastric varices. How do we explain at this point the high incidence of simultaneous gastric and esophageal varices? A possible explanation could be supplied by some recent studies on portal hemodynamics [19]. Such studies seem to demonstrate that in patients with simultaneous gastric and esophageal varices, the whole or part of the blood going out of the superior mesenteric vein can flow retrogradely into the splenic vein, suppling gastric varices through short and posterior gastric veins. In our series only 2.5% of patients with esophageal variceal bleeding had small-caliber varices. There are many methods available for grading esophageal varices, but perhaps the one adopted by our center, similar to that adopted by the Japanese Society for Portal Hypertension, has the advantage of being simple and facilitates comparison between different operators. We considered the presence of alternating edematous and hyperemic areas in the gastric antrum and body as indicative of congestive gastropathy, in accordance with observations of other authors [lo]. The incidence of congestive gastropathy in our patients was similar to that ob-

78 Table 2. Review of the literatureof sites of bleeding in patients with vafices References

Patients Varices only Gastric erosions Gastric ulcer Duodenal ulcer Mallory-Weiss

Sutton [15]

Tabinian and Graham [16]

Buset et al. [3]

Present study

222 65% 15% 8% 3% -

127 79.5% 8% 7% 2.3% 3%

165 54.5% 18.2% 10.9% 7.8% 3%

112 72.4% 14.3% 4.4% 8.9% -

served by McCormack and coworkers [8]. In our experience gastropathy is more frequent in patients with largecaliber varices, but the difference of incidence did not prove statistically significant. This finding would agree with the report of Adachi et al. [1], who observed that the increase in portal pressure is associated with an increase in vascular mucosal abnormalities. A number of authors [4, 13, 17] have clarified the pathogenesis of gastropathy. Thus, in spite of the fact that in cirrhothic patients there is an increase in blood flow towards the gastric wall, there is a reduction of blood flow to the mucosal layer because of edema and arteriovenous shunts in the submucosal layer. The defense capabilities of the mucosal layer are therefore reduced [1] and are more likely to be the possibility of damaged from exogenous agents. Management of such patients is difficult, and gastropathy is affected only by surgical techniques that reduce portal hypertension. In our analysis the incidence of esophagitis was not significantly different in patients who had already had a hemorrhage compared to those who had never previously bled. This finding is similar to that reported by Heil et al. [5], but is in contrast with the observations of Beppu and coworkers [2] who considered esophagitis a risk factor for bleeding in esophageal varices. In our series of 313 patients in whom endoscopy was delayed for up to 1 week after bleeding, 33.5% of the cases had a different gastrointestinal pathology that could have been the source of hemorrhage. If endoscopy is performed more than 24 h after the hemorrhage, definite detection of the bleeding source can be obtained only in 57.2% of cases. This figure increases to 98.2% if the examination is performed in the first 24 h. In this group we observed that in the majority of cases varices were the cause of the bleeding. These data are in accordance with the findings of Sutton [15] and Tabinian and Graham [16], but are significantly greater than those observed by Buset et al. [3] (Table 2). In 2 of our cases (1.8%), we obtained the diagnosis of variceal bleeding only by excluding other pathologies. Other authors have reported slightly higher percentages of diagnosis by using exclusion criteria [3, 14]. In patients with variceal hemorrhage, the varices were still bleeding at the time of the endoscopy in only one-third of the cases. In the remaining two-thirds, w~ observed only clots and fibrinoid aggregates on the varices. However, it should be remembered that the presence of a clot on the varix is not always a sign of variceal bleeding, but could represent gastroesophageal reflux of blood. Our experience indicates that a diligent search for other causes of upper

gastrointestinal hemorrhage should be undertaken in patients with and without variceal bleeding. This is particularly important in the presence of small varices which, in our experience, are responsible for only 2.5% of the total variceal bleeding. We did not encounter duodenal varices, which have been reported by others [3] as a potential source of bleeding. Erosive gastritis was in our experience the secondmost-frequent cause (after varices) responsible for hemorrhage in cirrhotic patients. Rector and Reynolds [12] also observed this, although these authors used the broader term of "hemorrhagic gastritis," which also includes acute ulceration. This high incidence of erosive gastritis can be explained by the presence of gastric microvasculature abnormalities, which predispose to mucosal injury (especially from alcohol in cirrhotic patients) [13]. Quintero et al. [11] described a capillary ectasia of the lamina propria in cirrhotic patients as being responsible for occult bleeding, leading to chronic anemia. It is necessary, therefore, to search for such alterations even in patients with signs of variceal bleeding. In conclusion, when an endoscopic examination is performed in patients with portal hypertension, the presence of esophageal varices must stimulate a careful search for a whole series of other pathologies, which occur frequently in these patients. The presence or the absence of different pathologies indicate different therapeutic choices. Furthermore, endoscopy in patients with upper gastrointestinal hemorrhage must be performed as soon as possible, and the endoscopist must exclude other potential sources of bleeding even in the presence of bleeding varices.

References 1. Adachi H, Maruyama M, Naito H (1986) Study of gastric mucosal chances and mucosal hemodynamicsin patients with liver cirrhosis. Dig Dis Sci 31 [Suppl 10]: 70-75 2. Beppu K, Inokuchi K, KoyanagiN, Nakayama S, Sakata H, Kitano S, KobayashiM (1981) Predictionof varicealhemorrhageby esophageal endoscopy.GastrointestEndosc 27:213 - 218 3. Buset M, Des Marez B, Baize M, Bourgeois N, Cremer M (1987) Bleeding esophagogastric varices: an endoscopic study. Am J Gastroentero182:241-244 4. HashizumeM, Tanaka K, InokuchiK (1983) Morphologyof gastric microcirculationin cirrhosis. Hepathology3:1008 - 1012 5. Heil T, Mattes P, LocprechtH (1980) Gastro-oesophagealreflux: an aetiologicalfactor for bleeding in oesophagealvarices?Br J Surg 67: 467-468

79 6. Hosking SW, Johnson AG (1988) Gastric varices: a proposed classification leading to management. Br J Surg 75: 195-196 7. McCormack Tr, Rose JD, Smith PM, Johnson AG (1983) Perforating veins and blood flow in oesophageal varices. Lancet II: 1442- 1444 8. McCormack TT, Sims J, Eyre-Brooke I (1985) Gastric lesions in portal hypertension: inflammatory or congestive gastropathy? Gut 26:1226-1232 9. Okuda K, Yasumoto M, Goto A, Kunisaki T (1973) Endoscopic observations of gastric varices. Am J Gastroentero160: 357- 365 10. Papazian A, Braillon A, Dupas JL (1986) Portal hypertensive gastric mucosa: an endoscopic study. Gut 27:1199-1203 11. Quintero E, Pique JM, Bompi JA, Bordas JM, Sentis J, Montserrat E, Bosch J, Rodes J (1987) Gastric mucosal vascular ectasias causing bleeding in cirrhosis. Gastroenterology 93: 1054-1061 12. Rector WJ, Reynolds TB (1985) Risk factors for hemorrhage from esophageal varices and acute gastric erosion. Gastroenterology 14: 139-153 13. Sarfeh IJ, Tarnawski A, Malki A, Mason JR, Mach T, Ivey KJ (1983) Portal hypertension and gastric mucosal injury in rats: effects of alcohol. Gastroenterology 84:987-993

14. Sivak MS (1988) Sclerotherapy of esophageal varices. Curt Ther Gastroenterol Liver Dis 6: 4 8 - 54 15. Sutton FM (1987) Upper gastrointestinal bleeding in patients with esophageal varices. What is the most common source? Am J Med 83: 273-275 16. Tabinian N, Graham DY (1987) Source of upper gastrointestinal bleeding in patients with esophageal varices seen at endoscopy. J Clin Gastroenterol 9: 279-282 17. Tarnawski A, Sarfeh IJ, Stachura J, Haiduczek A, Bui HX (1987) Portal hypertensive gastropathy or vasculopathy? Gastroenterology 3: 1008-1012 18. Trudeau W, Prindeville T (1983) Endoscopic injection sclerosis of bleeding gastric varices (abstract). Gastroenterology 84:1338 19. Vianna A, Hayes PC, Moscoso G, Drivers M, Portmann B, Westaby D, Williams R (1987) Normal venous circulation of the gastroesophageal junction. A route to understanding varices. Gastroenterology 93:876-889 20. Watanabe K, Kimura K, Matsutani S, Ohto M, Okuda K (1988) Portal hemodynamics in patients with gastric varices. Gastroenterology 95:434-440