Gastric mucosal responses to intrahepatic portosystemic shunting in patients with cirrhosis

GASTROENTEROLOGY 2000;118:905–911

LIVER, PANCREAS, AND BILIARY TRACT Gastric Mucosal Responses to Intrahepatic Portosystemic Shunting in Patients With Cirrhosis PATRICK S. KAMATH,* MARCO LACERDA,* DAVID A. AHLQUIST,* MICHAEL A. MCKUSICK,‡ JAMES C. ANDREWS,‡ and DAVID A. NAGORNEY§ *Division of Gastroenterology and Hepatology, Departments of ‡Diagnostic Radiology and §Surgery, Mayo Clinic, Rochester, Minnesota

Background & Aims: The response of gastric mucosal lesions in cirrhotic patients with portal hypertension, namely, portal hypertensive gastropathy (PHG) and gastric vascular ectasia (GVE), to transjugular intrahepatic portosystemic shunts (TIPS) is not known. Methods: Clinical and laboratory evaluation, upper gastrointestinal endoscopy, and Doppler ultrasonography were performed before placement of TIPS and 6 weeks, 3 months, and 6 months after TIPS in 54 patients. Thirty patients had mild PHG, 10 had severe PHG, and 14 had GVE. Results: Approximately 75% of the patients with severe PHG responded to TIPS as shown by improvement in endoscopic findings and by a decrease in transfusion requirements; 89% of patients with mild PHG had endoscopic resolution. Patients with GVE had neither endoscopic resolution nor a decrease in transfusion requirements after TIPS. There was no difference in mortality between the 2 groups. Conclusions: The results support the position that severe PHG and GVE may be different lesions. Mild and severe PHG respond to TIPS. Because GVE does not respond to TIPS, we recommend that TIPS be avoided for the treatment of gastrointestinal bleeding associated with GVE.

astric mucosal lesions in patients with portal hypertension, as seen on upper gastrointestinal endoscopy, include portal hypertensive gastropathy (PHG) and gastric vascular ectasia (GVE). In their severe form, these lesions may cause recurrent gastrointestinal bleeding. PHG occurs in up to 70% of patients with portal hypertension.1 Mild PHG is characterized by a mosaic appearance of the mucosa of the stomach. Severe PHG is characterized by discrete red spots, in association with a mosaic appearance of the background mucosa.2 GVE is a more recently recognized entity. This lesion is characterized by aggregates of red spots. When these aggregates are arranged in a linear pattern in the antrum of the stomach, the term gastric antral vascular ectasia (GAVE),3 or ‘‘watermelon stomach,’’ is used. The ectatic red spots

G

may be more diffuse and involve the proximal stomach, when they are termed as the diffuse variety of GVE. When red spots are seen in the stomach on endoscopy it may be difficult to differentiate severe PHG from GVE.4 If the background mucosa takes up a mosaic appearance and red spots are present within the mosaic mucosa, the term most often used to describe the changes is severe PHG. However, if the background mucosa is normal and without a mosaic appearance, the term used to describe the red spots is GVE. When it is difficult to differentiate on endoscopy between severe PHG and GVE, one might have to resort to gastric mucosal biopsy. The presence of ectatic blood vessels with fibrinous thrombi is reported to be more suggestive of vascular ectasia.5 It is important to emphasize that although some investigators tend to separate PHG from GVE,6 others consider it part of the same spectrum.7 PHG may be treated by the use of b-adrenergic–blocking agents, such as propranolol, which have been shown to decrease the risk of gastrointestinal bleeding.8 GVE, on the other hand, may be treated with thermal ablation,9 although experience with this therapy is limited in the setting of cirrhosis. There are several unanswered questions regarding GVE. It is uncertain whether GVE is part of the PHG spectrum or whether it is a distinct entity. The ideal therapy for GVE is also unknown. Potential therapeutic approaches in patients with gastrointestinal blood loss because of PHG or GVE could include pharmacological therapy with b-adrenergic blockers or estrogen/progesterone combinations, endoscopic thermocoagulation, surgically created portosystemic shunts, transjugular intrahepatic portosystemic shunts (TIPS), or liver transplantation. Abbreviations used in this paper: GAVE, gastric antral vascular ectasia; GVE, gastric vascular ectasia; PHG, portal hypertensive gastropathy; TIPS, transjugular intrahepatic portosystemic shunt. r 2000 by the American Gastroenterological Association 0016-5085/00/$10.00 doi:10.1053/gg.2000.7040

906

KAMATH ET AL.

GASTROENTEROLOGY Vol. 118, No. 5

However, the relative merits of these therapies are uncertain. Finally, it is unclear whether the response of GVE lesions to TIPS is similar to that of PHG. The aim of this study was to determine the therapeutic response to TIPS in patients with PHG and GVE using endoscopic changes as a primary endpoint, and post-TIPS hemoglobin level, transfusion requirements, mortality, and newonset hepatic encephalopathy as secondary endpoints.

Patients and Methods Patients Consecutive patients with gastric varices and mild PHG (n 5 30), severe PHG (n 5 10), and GVE (n 5 14) who underwent TIPS placement for prevention of recurrent gastrointestinal bleeding were studied prospectively. All patients had clinical, biochemical, and ultrasonographic or histological evidence of cirrhosis of the liver. Significant cardiopulmonary or renal comorbidity was absent in these patients. Clinical and laboratory evaluation, upper gastrointestinal endoscopy, and Doppler ultrasonography to evaluate shunt patency were carried out before placement of TIPS, and 6 weeks, 3 months, and 6 months after TIPS. Additionally, the patients had hemoglobin, white cell, and platelet counts checked at weekly intervals after TIPS for the first 6 weeks and whenever they had evidence of gastrointestinal bleeding. Patients with severe PHG and GVE were transfusion dependent, requiring red cell transfusion at a frequency of approximately 7–10 days before TIPS. Patients with GVE had discontinued endoscopic therapy at least 3 weeks before TIPS. Pharmacological therapy with propranolol in both the GVE and PHG groups was discontinued 48–72 hours before TIPS. Of the 14 patients with GVE lesions, 6 had GAVE lesions (watermelon stomach) and 8 had the diffuse variety of GVE. None of the 14 GVE patients had undergone endoscopic variceal sclerotherapy or variceal endoligation. All 6 patients with GAVE lesions had responded poorly to endoscopic therapy (neodynium yttrium-aluminum-garnet laser, n 5 4; argon plasma coagulator, n 5 2); none of the patients with diffuse GVE had received endoscopic thermocoagulation therapy. Of note, when the patients with GAVE were referred for the TIPS placement, the ‘‘watermelon’’ appearance was not obvious because endoscopic therapy disrupted the linear aggregates. The endoscopic examinations were stored on videotape in all 54 patients. Patient characteristics and hepatic hemodynamics are shown in Table 1. Oral iron supplementation was continued in all patients after the procedure. TIPS was performed using

standard techniques.10 The aim of the TIPS was to reduce the portacaval pressure gradient below 12 mm Hg or as close to 12 mm Hg as possible using a single Wallstent (Schneider, Minneapolis, MN). TIPS was carried out only if the hemoglobin level was greater than 8 g/dL; patients received transfusions if needed to increase hemoglobin above this level.

Endoscopic Criteria for PHG and GVE Red spots arranged in a linear pattern in the antrum without background mosaic appearance were designated as GAVE (Figure 1). Diffuse red spots either in the antrum, fundus, or body were designated as diffuse GVE (Figure 2). Diffuse GVE did not involve the body or fundus of the stomach in the absence of antral involvement. PHG was described using standard criteria.11 In brief, a background mosaic appearance of the mucosa without diffuse red spots was diagnosed as mild PHG. A background mosaic pattern with diffuse red spots was diagnosed as severe PHG (Figure 3). These changes of PHG always involved the fundus of the stomach and, to a variable extent, the distal stomach. No gastric biopsies were carried out, either before or after TIPS. Thus, the distinction between PHG and GVE was based purely on subjective endoscopic criteria.

Endoscopic Resolution In all patients, endoscopic findings were recorded when hemoglobin level was at least 8.0 g/dL and ultrasonography had confirmed patency of the shunts. Endoscopic resolution was defined as follows: for mild PHG, the mosaic appearance had to disappear; for severe PHG and GVE, the red spots had to disappear or almost completely disappear. The videotapes and images on each patient were reviewed independently by 2 of the investigators (P.S.K. and D.A.A.), each with more than 15 years of experience with endoscopy in patients with portal hypertension. Endoscopic criteria for diagnoses were agreed on before evaluation of the images. The 2 endoscopists were blinded as to the indication for TIPS as well as to the timing of endoscopy after TIPS. Forty-one images in 32 of the patients were reviewed on 2 occasions 3 weeks apart by the endoscopists and the endoscopic diagnosis made. These diagnoses were compared by a statistician to determine intraobserver and interobserver variation.

Transfusion Requirements The number of units of red cells that the patients required to keep the hemoglobin above 8.0 g/dL was recorded. This was expressed as the transfusion index, or the number of units transfused plus one divided by the months of follow-up, analogous to the bleeding index outlined at the Baveno II

Table 1. Patient Characteristics

Patients (n) Age ( yr) Child–Pugh score Pre-TIPS portocaval gradient (mm Hg ) Post-TIPS portocaval gradient (mm Hg )

Mild PHG

Severe PHG

GVE

P values

30 51 6 8.2 11.6 6 3.1 20.4 6 5.9 12.4 6 4.6

10 63.4 6 10.6 11.2 6 4.2 21.4 6 6.4 13.2 6 5.9

14 68.4 6 4.8 8.7 6 2.2 18.6 6 7.8 12.1 6 6.7

0.04, GVE vs. mild PHG 0.05, GVE vs. PHG NS NS

May 2000

TIPS FOR GASTRIC VASCULAR ECTASIA

907

Figure 1. Endoscopic appearance of the stomach in a patient with GAVE before laser therapy. Note the aggregates of red spots arranged in a linear pattern and the absence of a background mosaic pattern.

Figure 2. Endoscopic appearance of the stomach in a patient with diffuse variety of GVE. Note aggregates of red spots without a significant mosaic background.

Figure 3. Endoscopic appearance of the stomach in severe PHG. Note the mosaic background with red spots.

908

KAMATH ET AL.

conference.12 Although ferritin level and relative distribution width of red cells were measured in all patients, they were not included in the analyses because they are unreliable tests in the presence of frequent red cell transfusions.13 Hemoglobin 2 weeks after TIPS was compared with the hemoglobin immediately before TIPS in all surviving patients. Although hemoglobin values were available beyond that period, the requirement for red cell transfusion in some patients compromised the analysis. No patient received a transfusion within 2 weeks of TIPS.

Mortality Mortality due to any cause was also recorded and separated into 30-day mortality and 1-year mortality. Patients were followed up to death (n 5 27), liver transplantation (n 5 3), or to at least 1 year (n 5 24).

Hepatic Encephalopathy Hepatic encephalopathy was graded using standard criteria.14 Worsening encephalopathy or new-onset hepatic encephalopathy over a 1-year period was recorded. Lactulose was administered only to patients with overt encephalopathy.

Evaluation of Shunt Patency A midshunt velocity less than 50 cm/s on Doppler ultrasonography was considered as a criterion for diagnosis of shunt stenosis,10 and hepatic venography was carried out. No endoscopic investigations were routinely done in these patients beyond 6 months after TIPS; however, ultrasonography to determine shunt patency was continued at 3-month intervals. Shunt revisions were required in 3 patients in the mild PHG group and in none in the severe PHG and GVE groups.

Statistical Analysis Comparison within groups (mild PHG, severe PHG, and GVE) for outcomes after TIPS was made using the Student t test for paired samples. Differences between the 3 groups were determined using analysis of variance. A P value of ,0.05 was determined to be significant. Intraobserver and interobserver agreement were analyzed using the Cohen k statistic.

GASTROENTEROLOGY Vol. 118, No. 5

Table 2. Endoscopic Findings After TIPS Endoscopic resolution after TIPS

PHG Mild (n 5 30) Severe (n 5 10) GVE (n 5 14)

6 Weeks

3 Months

6 Months

18/28 (64%) 3/9 (33%) 0/10 (0%)

23/27 (85%) a 6/8 (75%) 1/8 (12.5%) b

24/27 (89%) a 5/7 (71%) 1/8 (12.5%) b

NOTE. Mortality in the various groups was as follows: mild PHG, 2 patients died at 6 weeks, 3 at 3 months, and 3 at 6 months; severe PHG, 1 at 6 weeks, 2 at 3 months, and 3 at 6 months; and GVE, 4 at 6 weeks, 6 at 3 months, and 6 at 6 months. aP 5 0.01, mild PHG vs. GVE. bP 5 0.025, severe PHG vs. GVE.

discussion between the observers. The other disagreements were on follow-up endoscopies between mild PHG and normal in patients who initially had severe PHG (n 5 3). The earliest change, as seen on endoscopy, was with mild PHG with resolution in most patients at 6 weeks, and in approximately 85% of patients at 3 and 6 months. There was slow initial response to severe PHG, but, by 3 months, more than two thirds of the patients had resolution in PHG lesions, which stayed resolved at 6 months. In sharp contrast, there was no resolution in GVE lesions at 6 weeks and a poor response at 3 and 6 months. The only patient with endoscopic changes labeled as GVE that resolved had diffuse lesions. This was the patient in whom there was a disagreement between the observers as to the diagnosis, GVE or severe PHG. In 1 patient with mild PHG in whom esophageal and gastric varices had completely regressed at 6 months, there was new onset of vascular ectatic lesions in the antrum and fundus of the stomach. Two patients with severe PHG and 1 with GVE underwent liver transplantation with complete resolution of gastrointestinal bleeding. In the absence of gastrointestinal bleeding after liver Table 3. Results After TIPS Follow-up

Results The endoscopic resolution of PHG, mild as well as severe, and GVE at 6 weeks, 3 months, and 6 months after TIPS is shown in Table 2. Mortality, transfusion requirements, and new-onset hepatic encephalopathy are shown in Table 3. Endoscopic Changes The k score for interobserver agreement for endoscopic changes was 0.8, and the intraobserver agreement was 0.955 (P.S.K.) and 0.922 (D.A.A.). The only patient in whom there was disagreement as to whether this was severe PHG or GVE was labeled as GVE after a

Follow-up (mo) Mortality 30 days 1 yr RBC transfusion index (U ) Pre-TIPS Post-TIPS New-onset hepatic encephalopathy 1 yr

Mild PHG

Severe PHG

GVE

11.2 6 9.4

17.4 6 6.2

13.8 6 7.1

2/30 16/30

1/10 4/10

3/14 7/14

a

14/20

2.9 6 2.0 0.6 6 0.8 b

4/9

3.8 6 2.8 3.4 6 1.9

6/7

RBC, red blood cell. had gastric varices and presented with acute gastrointestinal bleeding rather than chronic blood loss. bP 5 0.04 vs. pre-TIPS; P 5 0.02 vs. post-TIPS GVE. aPatients

May 2000

transplantation, an endoscopic examination could not be justified. Transfusion Requirements There was a significantly lesser requirement for red cell transfusions after TIPS in the severe PHG group compared with the GVE group as determined by the transfusion index (0.6 6 0.8 vs. 3.4 6 1.9; P 5 0.02). No significant decrease in transfusion requirements was found in the GVE group after TIPS compared with before TIPS (Table 3). A significant reduction in hemoglobin level was observed after TIPS in the GVE group (before TIPS vs. after TIPS; P 5 0.02), but no significant change in the mild PHG and severe PHG groups (P 5 0.07) (Figure 4). By 1-way analysis of variance, hemoglobin level after TIPS was significantly different among the 3 groups (P , 0.05). Mortality The mean follow-up of patients was more than 1 year with no significant difference in mortality at 30 days and at 1 year among the mild PHG, severe PHG, and GVE groups (Table 3). All deaths were attributed to liver failure. The patients who died were not candidates for liver transplantation because of their advanced age. Hepatic Encephalopathy New-onset hepatic encephalopathy at the end of 1 year was almost invariable in the patients with GVE (Table 3). Patients with severe PHG had less hepatic encephalopathy than the mild PHG or GVE groups, but the differences did not reach statistical significance (P 5 0.06).

Figure 4. Scatter plot of hemoglobin before and after TIPS in patients with mild PHG, severe PHG, and GVE. Hemoglobin after TIPS was significantly lower than before TIPS in the GVE group (P , 0.02). There was no significant change in the mild and severe PHG groups (P 5 0.07).

TIPS FOR GASTRIC VASCULAR ECTASIA

909

Discussion The results of this study show that PHG often improves after TIPS, but GVE may not improve. The response to TIPS is usually observed within 2 weeks in the PHG group with stabilization of hemoglobin level and decrease in transfusion requirements. GVE lesions occur in patients with relatively preserved liver function, but encephalopathy and mortality are high in these patients after TIPS, despite their relatively preserved liver function before TIPS. The results of this study, which is descriptive in nature, suggest that GVE lesions may be distinct from PHG lesions. It should be emphasized that the distinction between PHG and GVE was purely subjective because no histological examination was performed. We cannot rule out the possibility that the 2 patients with severe PHG who did not respond to TIPS at 6 months may have had GVE or that the patient labeled as having GVE who responded to TIPS may have actually had severe PHG. Given the poor response in our series of GVE lesions to TIPS, compared with the relatively favorable response of severe PHG, it is important that these lesions be distinguished. Two important issues need to be addressed further: the pathogenesis of the gastric lesions in portal hypertension and the most appropriate treatment of these conditions. The mechanism by which red spots and vasodilatation of the gastric submucosal vessels occur remains unknown. Intermittent obstruction of submucosal vessels has been hypothesized as a factor contributing to vascular ectasia.4 As our study shows, portal hypertension is not a prerequisite for the development of GVE, but liver transplantation seems to correct the defect. Neither acid secretion15 nor Helicobacter pylori16 seem to play a major role. Portal hypertension is a prerequisite for development of PHG

910

KAMATH ET AL.

and, thus, impairment of gastric venous drainage caused by portal hypertension may be important. However, portal hypertension may not be the only pathogenetic factor, because there does not seem to be any direct correlation between the severity of PHG and the hepatic venous pressure gradient.17 Several investigators have shown an increased blood flow in the gastric mucosa of portal hypertensive animals.18,19 Studies in humans using laser Doppler flowmetry and reflectance spectrophotometry have shown both increased gastric perfusion20 or decreased perfusion.21 The difficulty in interpreting these findings is compounded by the fact that red cell content in the gastric mucosa influences blood flow, as assessed by laser Doppler or reflectance spectrophotometry; anemia tends to result in overestimation of the mucosal flow.22 Our findings show that PHG has a reasonably good response to TIPS, indicating that the dominant pathogenetic mechanism in PHG is probably high portal pressure. On the other hand, reduction of portal pressure does not seem to significantly benefit GVE lesions. We hypothesize that the GVE lesions may be related to increased mucosal flow and splanchnic vasodilatation, which is known to occur in patients with chronic liver disease.23 The worsening of the GVE lesions after TIPS may be related to the increase in the cardiac output that occurs after TIPS.24,25 The second important issue is determination of the most appropriate treatment for patients with portal hypertension and gastric mucosal lesions. PHG shows a favorable response to b-adrenergic–blocking agents,8 and we believe this group of drugs should be tried in all patients with symptomatic PHG. Although mild PHG responds to TIPS, significant blood loss is rare in these patients and any therapy seldom indicated. When treatment of severe PHG with pharmacological agents fails, TIPS would be a viable option. Surgical shunts have also been used with success in the treatment of severe PHG.26 The appropriate treatment for GVE lesions is still unknown. Our data corroborate findings of a recent study that TIPS does not prevent bleeding from GVE.6 Iron supplementation may be sufficient in patients with GVE lesions whose bleeding is mild. This may be combined with estrogen/progesterone combinations27 as a means of preventing further bleeding. Endoscopic coagulation may be effective in GAVE lesions (watermelon stomach), but, in the diffuse form, results are poor.9 When these treatments fail, gastric resection can be considered. Antrectomy is effective,6 but surgical mortality and morbidity may be substantial in the setting of cirrhosis. The results of the present study and other reports suggest that TIPS or surgical portacaval shunts28 are contraindicated. Further, because patients with GVE lesions typi-

GASTROENTEROLOGY Vol. 118, No. 5

cally are older, they are at a higher risk for development of postportosystemic shunt hepatic encephalopathy.29 We thus conclude that both PHG and GVE lesions occur in cirrhotic patients with portal hypertension. However, only PHG seems to respond to portal decompression by TIPS. Because GVE lesions do not respond to TIPS, the present study supports the position that GVE and severe PHG have separate pathogenic mechanisms. We recommend that TIPS be avoided as treatment of gastrointestinal bleeding associated with GVE.

References 1. Pique JM. Portal hypertensive gastropathy. Baillieres Clin Gastroenterol 1997;11:257–270. 2. McCormack TT, Sims J, Eyre-Brook I, Kennedy H, Goepel J, Johnson AG, Triger DR. Gastric lesions in portal hypertension: inflammatory gastritis or congestive gastropathy? Gut 1995;26: 1226–1232. 3. Jabbari M, Cherry R, Lough JO, Daly DS, Kinnear DG, Goresky CA. Gastric antral vascular ectasia: the watermelon stomach. Gastroenterology 1984;87:1165–1170. 4. Cale`s P, Zabotto B, Meskens C, Caucanas JP, Vinel JP, Desmorat H, Fermanian J, Pascal JP. Gastroesophageal endoscopic features in cirrhosis: observer variability, interassociations, and relationship to hepatic dysfunction. Gastroenterology 1990;98: 156–162. 5. Payen JL, Cale`s P, Voigt JJ, Barbe S, Pilette C, Dubuisson L, Desmorat H, Vinel JP, Kervran A, Chayvialle JA, Pascal JP. Severe portal hypertensive gastropathy and antral vascular ectasia are distinct entities in patients with cirrhosis. Gastroenterology 1995; 108:138–144. 6. Spahr L, Villeneuve JP, Dufresne MP, Tasse´ D, Bui B, Willems B, Fenyes D, Pomier-Layragnes G. Gastric antral vascular ectasia in cirrhotic patients: absence of relation with portal hypertension. Gut 1999;44:739–742. 7. Viggiano TR, Gostout CJ. Portal hypertensive intestinal vasculopathy. A review of the clinical, endoscopic, and histopathologic features. Am J Gastroenterol 1992;87:944–954. 8. Perez Ayuso RM, Pique JM, Bosch J, Panes J, Gonzales A, Perez R, Rigan J, Quintero E, Valderrama R, Viver J, Rodes J. Propranolol in the prevention of rebleeding from portal hypertensive gastropathy. Lancet 1991;337:1431–1434. 9. Gostout CJ, Ahlquist DA, Radford CM, Viggiano TR, Bowyer BA, Balm RK. Endoscopic laser therapy for watermelon stomach. Gastroenterology 1989;96:1462–1465. 10. Kamath PS, McKusick MA. Transjugular intrahepatic portosystemic shunts. Baillieres Clin Gastroenterol 1997;11:327–349. 11. Paulo Spina G, Arcidiacono R, Bosch J, Pagliaro L, Burroughs AK, Santambrogio R, Rossi A. Gastric endoscopic features in portal hypertension: final report of a consensus conference, Milan, Italy, September 19, 1992. J Hepatol 1994;21:461–467. 12. de Franchis R, ed. Portal hypertension II. Proceedings of the Second Baveno International Consensus Workshop on Definitions, Methodology and Therapeutic Strategies. Cambridge, MA: Blackwell Scientific, 1996:10–17. 13. Lee GR. Anemia: a diagnostic strategy. In: Lee GR, ed. Winthrop’s clinical hematology. 10th ed. Baltimore: Lippincott, Williams & Wilkins, 1999:908–933. 14. Conn HO, Leevy CM, Vlahcevic ZR, Rodgers JB, Maddrey WC, Saeff, L, Levy LL. Comparison of lactulose and neomycin in the treatment of chronic portal systemic encephalopathy. A double blind controlled trial. Gastroenterology 1977;72:573–583. 15. Quintero E, Pique JM, Bombi JA, Bordas JM, Sentis J, Elena M, Bosch J, Rodes J. Gastric vascular ectasia causing bleeding in

May 2000

16.

17.

18.

19.

20.

21.

22.

23.

24.

patients with cirrhosis: a distinct entity associated with hypergastrinemia and low serum levels of pepsinogen. Gastroenterology 1987;93:1054–1061. McCormick PA, Sankey EA, Cardin F, Dhillon AP, McIntyre N, Burroughs AK. Congestive gastropathy and Helicobacter pylori: an endoscopic and morphometric study. Gut 1991;32:351–354. Vigneri S, Termini R, Piraino A, Scialabba A, Pisciotta G, Fontana N. Stomach and liver cirrhosis. Endoscopy, morphological and clinical correlations. Gastroenterology 1991;101:472–478. Benoit JN, Barrowman JA, Harper SL, Kvietys PR, Granger DN. Role of humoral factors in gastrointestinal hyperemia associated with chronic portal hypertension. Am J Physiol 1984;247:G486– G493. Geraghty JG, Anderson WJ, Carter DC. Autoradiographic study of the regional distribution of gastric blood flow in portal hypertension rats. Gastroenterology 1989;97:1108–1114. Pane´s J, Bordas J, Pique´ JM, Bosch J, Garcı´a-Paga´n JC, Feu F, Casadevall M, Tere´s J, Rode´s J. Increased gastric mucosal perfusion in cirrhotic patients with portal hypertensive gastropathy. Gastroenterology 1992;103:1875–1882. Iwao T, Toyonaga A, Ikegami M, Oho K, Sumino M, Harada M, Sakaki M. Reduced gastric mucosal blood flow in patients with portal hypertensive gastropathy. Hepatology 1993;18:36–40. Casadeval M, Panes J, Pique JM, Bosch J, Teres J, Rodes J. Limitations of laser Doppler and reflectance spectrophotometry in estimating gastric mucosal flow. Am J Physiol 1992;263:G810– G815. Bomzon A, Blendis LM. The nitric oxide hypothesis and the hyperdynamic circulation in cirrhosis. Hepatology 1994;20:1343– 1350. Azoulay D, Castaing D, Dennison A, Martino W, Eyraud D, Bismuth

TIPS FOR GASTRIC VASCULAR ECTASIA

25.

26.

27. 28.

29.

911

H. Transjugular intrahepatic portosystemic shunt worsens the hyperdynamic circulatory state of the cirrhotic patient: preliminary report of a prospective study. Hepatology 1994;19:129– 132. Lotterer E, Wengert A, Fleig WE. Transjugular intrahepatic portosystemic shunt: short-term and long-term effects on hepatic and systemic hemodynamics in patients with cirrhosis. Hepatology 1999;29:632–639. Orloff MJ, Orloff MS, Orloff S, Haynes KS. Treatment of bleeding from portal hypertensive gastropathy by portocaval shunt. Hepatology 1995;21:1011–1017. Manning R. Estrogen/progesterone treatment of diffuse antral vascular ectasia. Am J Gastroenterol 1995;90:154–156. Cale`s P, Voigt JJ, Payen JL, Bloom E, Berg P, Vinel JP, Pradere B, Broussy P, Pascal JP. Diffuse vascular ectasia of the antrum, duodenum and jejunum in a patient with nodular regenerative hyperplasia. Lack of response to portosystemic shunt or gastrectomy. Gut 1993;34:558–561. Sanyal AJ, Freedman AM, Shiffman ML, Purdum PP, Luketic VA, Cheatham AK. Portosystemic encephalopathy after transjugular intrahepatic portosystemic shunt: results of a prospective controlled study. Hepatology 1994;20:46–55.

Received August 4, 1999. Accepted January 6, 2000. Address requests for reprints to: Patrick S. Kamath, M.D., Mayo Clinic, 200 First Street S.W., Rochester, Minnesota 55905. e-mail: [email protected]; fax: (507) 284-0538. The authors thank Dr. Tousif Pasha for help with the statistics and Linda Veer for expert secretarial assistance.