Severe Portal Hypertensive Gastropathy and Antral Vascular Ectasia Are Distinct Entities in Patients With Cirrhosis

Transcription

1 GASTROENTEROLOGY 1995;108: LIVER, PANCREAS, AND BILIARY TRACT Severe Portal Hypertensive Gastropathy and Antral Vascular Ectasia Are Distinct Entities in Patients With Cirrhosis JEAN-LOUIS PAYEN,* PAUL CALI~S,* JEAN-JACQUES VOIGT, SOPHIE BARBE,* CHRISTOPHE PILETTE,* LILiANNE DUBUISSON, II HERVl~ DESMORAT,* JEAN-PIERRE VINEL,* ALAIN KERVRAN, ~ JEAN-ALAIN CHAYVIALLE, # and JEAN-PIERRE PASCAL* *service d'h6pato-gastroenterologie and d'anatomie Pathologique, Centre Hospitalier Universitaire Purpan, Toulouse; ~Service d'h6pato- Gastroent~rologie, Centre Hospitalier Universitaire, Angers; I[Laboratoire des Interactions Cellulaires, Universit6 Bordeaux II, Bordeaux; ~Unit~ Centre National de la Recherche Scientifique INSERM de Pharmacologie-Endocrinologie, Montpellier; and #INSERM Unit~ 45, H6pital Edouard Herriot, Lyon, France Background~Aims: Whereas severe portal hypertensive gastropathy and gastric antral vascular ectasia (GAVE) have been separately defined in patients with cirrhosis, there is much confusion in the literature because they are both characterized by red spots at endoscopy. This prospective study compared clinical, biochemical, and pathological features of these syndromes. Methods: Three groups of patients with cirrhosis and either GAVE (n = 14), severe portal hypertensive gastropathy (n = 14), or no gastric features at endoscopy (controls; n = 10) were included. Results: No difference was found between patients with gastropathy and controls. Patients with GAVE presented with the following significant differences compared with other patients: a higher Child-Pugh score, a lower blood level of hemoglobin and gastrin, and a higher intestinal blood loss. At pathological examination, these patients more frequently had vascular ectasia (P = 0.04), spindle cell proliferation (P < 0.01), fibrohyalinosis (P = 0.004), and Gilliam's score of ->2 (P < 0.05); thrombi were encountered only in patients with GAVE (P = 0.006). Using discriminant analysis, spindle cell proliferation and fibrohyalinosis were the only significant variables yielding a diagnostic accuracy of 85% for GAVE and gastropathy. Conclusions: GAVE and severe portal hypertensive gastropathy are two distinct entities. G astric mucosal lesions such as portal hypertensive (PHT) gastropathy are increasingly recognized at endoscopy as common features in patients with portal hypertension. 1'2 Furthermore, they are recognized as a cause of anemia 3-5 or even overt hemorrhage. 6'7 The most widely used classification of endoscopic aspects of PHT gastropathy or congestive gastropathy was established by McCormack et al. 8 In addition, in 1984, Jabbari et al. 9 described the gastric antral vascular ectasia (GAVE) syndrome that differed from severe PHT gastropathy by its endoscopic appearance (watermelon stomach), its occurrence in noncirrhotic patients in most cases, and also its microscopic alterations in the antral mucosa. 1 '11 However, in cirrhosis, this distinction has not yet been universally accepted, and some investigators include the GAVE syndrome in the gastric red spot pattern of severe PHT gastropathy ~2'13 or believe that a large overlap exists between GAVE syndrome and severe PHT gastropathy. ~4 This prospective study investigated whether any difference could be detected in the clinical, biochemical, and pathological features between severe PHT gastropathy and GAVE in patients with cirrhosis who were systematically screened at endoscopy for portal hypertension. Materials Patients and Methods Inclusion criteria for the study were that the patients presented with cirrhosis, were between 18 and 80 years of age, agreed to participate in the study, and had gastric patterns at time of endoscopy as described below. The diagnosis of cirrhosis was assessed by usual clinical and biochemical criteria or after a liver biopsy was performed. All patients underwent an upper gastrointestinal endoscopy (Olympus Evis 100; Japan) to screen portal hypertension, performed by a well-experienced senior endoscopist. They were separated into three groups according to the following endoscopic criteria: 1. The gastropathy group consisted of 14 patients with gastric mucosal red spots, especially in the gastric fundus and corpus. These lesions met the criteria of McCormack et al.* for severe PHT gastropathy: "discrete red spots analogous to the cherry red spots described in the esophagus." All of Abbreviations used in this paper: GAVE, gastric antral vascular ectasia; PHT, portal hypertensive by the American Gastroenterological Association /95/$3.00

2 January 1995 PORTAL HYPERTENSIVE GASTROPATHY AND ANTRAL VASCULAR ECTASIA 139 Table 1. Clinical and Biochemical Data According to the Three Groups of Patients Control Severe PHT gastropathy GAVE P (n = 10) (n = 14) (n = 14) (analysis of variance) Age (yr) 62 ± ± _,t 13 NS Sex (M/F) (5/5) (10/4) (12/2) NS Child-Pugh score 6.4 ± 1.4 6, ± 2.3 <0.05 a Cause of cirrhosis Alcohol NS b Virus Autoimmune hepatitis Primary biliary Unknown Esophageal varices NS Sclerotherapy <0.02 ~-blocker NS Blood loss (ml/day) ND 8.9 ± ± 35 <0.05 Basal acid output (meq/h) ND ± 0.7 NS Peak acid output (meq/h) ND 50.4 ± NS Hemoglobin level (g/dl) 13.4 _~ _ ± 2.4 <0,05 a Anemia c 1/10 3/14 6/14 NS Ferritin (l~g/ml) 280 ± ± ± 86 NS Glucagon (fmol/ml) 45 _ ± 7 60 ± 10 NS Serotonin (l~g/ml) 373 ± ± ± 70 NS Vasoactive intestinal peptide (pg/ml) 5.6 ± ± 0.8 8,9 _+ 2.1 NS Gastrin (pg/ml) 98 ± ± <0.05 a NS, not significant; ND, not done. agave vs. PHT gastropathy or control; P < baicohol vs. other causes. CAnemia, hemoglobin < 11.5 g/de these patients also had a mosaic pattern. Patients with "diffuse haemorragic gastritis" patterns were not included. 2. The GAVE group consisted of 14 patients with red spots in the gastric antrum, either aggregated in linear stripes converging on the pylorus as described by Jabbari et al. (watermelon stomach, n = 7) or diffusely spread in the antrum as described by Lee et al. (diffuse form, n = 7) as GAVE. 9a4 3. The control group consisted of 10 patients without any gastric abnormalities at endoscopy. In particular, no mosaic pattern was present. Criteria of noninclusion were active gastrointestinal bleeding; administration of potentially ulcerogenic drugs, antacids, and H2-receptor antagonists; or alcohol intake 15 days before inclusion. Patients gave oral informed consent for the protocol, which was approved by the local ethics committee. Methods For each patient, the following parameters were recorded: age, sex, Child-Pugh score, ~5 blood hemoglobin content, and ferritinemia. Digestive blood losses were assessed over a 5-day period by 51 chromium-labeled red cells. Previous history of endoscopic sclerotherapy or B-blocker use was assessed for the patients with GAVE or PHT gastropathy (Table 1). Gastric acid secretion (basal acid output and peak acid output) was measured by gastric intubation before and after stimulation by pentagastrin (6 ~tg/kg of body weight; ICI Pharma, Cergy, France). Gastric biopsy specimens of red spots were obtained using Olympus FB 19 K forceps, two in the corpus and two in the antrum. Biopsy specimens were obtained from red spots when they were present. Biopsy specimens were fixed in ethanolbased Bouin's fluid (Duboscq-Brasil) and embedded in paraffin. In all cases, three step sections stained with H&E were available. Examination was performed by the same pathologist who was unaware of the specimen's group. Morphological alterations were recorded using Gilliam's classification, u~ which includes the following two criteria: (1) evaluation of mucosal vessels for the presence of fibrin thrombi and/or ectasia, which is scored as 0 if these anomalies are absent, 1 if thrombi or ectasia are present, and 2 if thrombi and ectasia are present; and (2) spindle cell proliferation (smooth muscle cell and myofibroblast hyperplasia) into superficial mucosa, which is scored as 0 if absent, 1 if increased, and 2 if markedly increased. The Gilliam's score also ranges from 0 to 4. Fibrohyalinosis was also recorded as present or absent. Fibrohyalinosis is a homogeneous substance, light pink on H&E staining, around the ectatic capillaries of the lamina propria. Moreover, inflammation was recorded as "dense" or "absent or mild". The plasma levels of the following hormones were measured: glucagon (radioimmunoassay) according to a method described elsewhere, ~6 serotonin (overload fluorometry), gastrin (radioimmunoassay: GASK-PR; CIS bin international, Gif-sur-Yvette, France), and vasoactive intestinal peptide (radioimmunoassay) according to a previously described method, iv

3 140 PAYEN ET AL. GASTROENTEROLOGY Vol. 108, No. 1 Table 2. Pathological Data According to the Three Groups of Patients Gastric antral vascular Lesion Location Control (%) PHT gastropathy (%) ectasia (%) P~ Thrombi Corpus Antrum (50) Ectasia Corpus 1 (10) 0 3 (21) NS Antrum 3 (30) 9 (64) 14 (100) 0.04 Spindle cell proliferation Corpus Antrum 0 4 (29) 12 (86) <0.01 Gilliam's score of ->2 Corpus (38) Antrum 0 3 (21) 11 (79) <0.05 Gilliam's score of ->3 Antrum (50) Fibrohyalinosis Corpus 1 (10) Antrum 1 (10) 5 (36) 13 (92) GAVE score of ->3 Antrum 0 1 (7) 10 (71) <0.01 GAVE score of ->4 Antrum 0 0 (0) 7 (50) 0.01 NS, not significant. acomparison between PHT gastropathy and GAVE groups. There was no significant difference between PHT gastropathy and control groups. Statistics Quantitative variables (expressed as mean + SD) were tested using Student's t test when distribution was normal, otherwise by the Mann-Whitney test; multiple comparisons were performed using variance analysis. Qualitative variables were compared by X 2 test or by Fisher's Exact Test when expected numbers were small. Tests were two tailed. Discriminant linear analysis was used for multivariate studies. Results From December 1990 to July 1991, 38 consecutive patients with cirrhosis, who were eligible, were included in the study. The cause of cirrhosis was alcohol in 65% of cases. None of the patients had evidence of gastric symptoms, associated autoimmune disorders, or active bleeding. Comparison of clinical and biochemical data (Table 1) between the three groups showed that patients with GAVE syndrome had a more advanced liver disease, as assessed by the Child-Pugh score, than patients in the PHT gastropathy and control groups. They also presented with lower blood hemoglobin levels and increased digestive blood loss. Gastric acid secretion and blood hormone levels did not differ between the three groups except for gastrinemia, which was significantly lower in the GAVE group than in the other two groups. There was no significant difference in histopathologic parameters between the PHT gastropathy and the control groups (Table 2). A comparison of the histopathologic parameters between the GAVE and PHT gastropathy groups showed no difference in the corpus (Table 2); therefore, the following data, especially those concerning scores, deal with the antrum. Vascular ectasia and spindle cell proliferation (fibromuscular hyperplasia) were each encountered in the antrum significantly more often in patients with GAVE than with PHT gastropathy (Table 2). Thrombi were never encountered in the severe PHT gastropathy group (Figures l and 2). Consequently, a Gilliam's score in the antrum of -->2 was more frequent in the GAVE group than in the PHT gastropathy group (79% vs. 21%, respectively; P < 0.05). Furthermore, fibrohyalinosis occurred more frequently in the GAVE group than in the PHT gastropathy group (92% vs. 36%, respectively; P = 0.004). Therefore, we calculated a new score that combined the Gilliam's score and fibrohyalinosis (range, 0-5) that we named the GAVE score (Table 2). A GAVE score of -->3 was more likely to distinguish GAVE from PHT gastropathy (71% vs. 7%, respectively) than a Gilliam's score of-->2 (79% vs. 21%, respectively). There was no difference when the cutoff was set at 4 for the GAVE score and 3 for the Gilliam's score (50% vs. 0, for both scores). Also, the difference of the mean GAVE score between GAVE and PHT gastropathy groups was more significant than that of the Gilliam's score (Table 3). Consequently, a GAVE score of -> 3 was the best indicator of the presence of GAVE. Only mild inflammation was sometimes observed in the groups. Using discriminant analysis in GAVE and PHT gastropathy groups, the four elementary histopathologic lesions yielded a diagnostic accuracy of 85% by stepwise regression; spindle cell proliferation was isolated at first step and fibrohyalinosis at second step (Table 4). The addition of significant (in univariate analysis) clinical and biochemical variables (i.e., Child-Pugh score, blood hemoglobin level, and gastrinemia) to the four lesions did not increase the diagnostic accuracy (data not shown). Because gastric red spots are located in different areas of the stomach in patients with GAVE or PHT gastropathy, it was interesting to compare pathological data between the corpus of patients with PHT gastropathy and the

4 Janua~ 1995 PORTAL HYPERTENSIVE GASTROPATHY AND ANTRAL VASCULAR ECTASIA 141 Table 3. Pathological Scores According to the Three Groups of Patients Score Location Gilliam Corpus Antrum GAVE Antrum PHT gastropathy GAVE P (analysis of variance) 0.1 ± _* _ 0.7 0, ± 1.1 NS < [ P < _ _+ 1,0 Control [ P<IO 2 J ± 1.1 I P<10-3 l P< 10-" < J l NS, not significant. Figure 1. Histopathologic appearance of antral mucosa in severe congestive gastropathy. Several mucosal capillaries (arrows) seem weakly ectatic (H&E; original magnification 250 ). antrum of patients with GAVE. It is noteworthy that the differences between the two groups were greater than when comparing identical areas (Tables 2 and 3). Discussion In this series, patients with GAVE syndrome differed from those with severe PHT gastropathy by having more advanced liver disease, as assessed by the ChildPugh score, and a lower blood hemoglobin level. The lower blood hemoglobin level in patients with GAVE syndrome than in patients with severe PHT gastropathy should be noted because the patients in this study were not suspected of digestive bleeding before endoscopy and none of the patients with GAVE had a previous history of overt gastrointestinal bleeding. This supports the literature because GAVE syndrome is often associated with anemia or overt hemorrhage, ~8'19 whereas even severe PHT gastropathy often occurs without anemia. In three studies, no relationship was found between blood hemo- Table 4. Discriminant Analysis in GAVE and PHT Gastropathy Groups Diagnostic accuracy Figure 2. Histopathologic appearance of antral mucosa in GAVE. Mucosal capillaries appear ectatic (E) with a fibrin thrombus (T). They are surrounded by fibrohyalinosis (H) with fibromuscular hyperplasia (F) (H&E; original magnification 200 ). Variable (%) Ectasia" Thrombi b Fibrohyalinosisa Spindle cell proliferation b Ectasia, thrombi, fibrohyalinosis, a spindle cell proliferation ~ Ectasia, thrombi, fibrohyalinosis, spindle cell proliferation with stepwise regression First step, spindle cell proliferation Second step, fibrohyalinosis Gilliam's score b GAVE score ~ NOTE. Different combinations of pathological variables that were significant in univariate analysis are presented. ap < op < 0.01

5 142 PAYEN ETAL. GASTROENTEROLOGY Vol. 108, No. 1 globin levels and the presence or grade of PHT gastropathy In that respect, digestive blood losses were significantly more marked in the GAVE syndrome group than in the PHT gastropathy group. Hormone level measurements did not help to distinguish GAVE syndrome from severe PHT gastropathy. Low levels of gastrinemia in the GAVE group were not expected and cannot be clearly explained. This finding does not support the association of GAVE syndrome, hypergastrinemia, and achlorhydria described by Quintero et al. 4 who suggested that these abnormalities could play a role in the pathogenesis of GAVE syndrome in patients with portal hypertension. There is a discrepancy in the data relevant to gastrinemia in patients with cirrhosis: levels of gastrinemia have been found to be increased, 23 decreased, 24 or normal 25 in basal conditions. In contrast, basal gastric secretion and maximal acid output were always within normal limits. 24'26 No relationship between PHT gastropathy and humoral factors, such as glucagon and catecholamines, 21 have been described so far. Moreover, no association between GAVE and plasma levels of glucagon, secretin, or norepinephrine have been reported. 2v This study showed more severe hepatic dysfunction in patients with GAVE than in patients with severe PHT gastropathy. In the literature, no relationship could be found between PHT gastropathy and the degree of hepatic dysfunction. 3'28'29 Patients with severe PHT gastropathy frequently had a previous history of endoscopic sclerotherapy, which is not surprising because sclerotherapy is a predictive factor of this condition. 5'3 Gastric mucosal biopsy specimens were studied according to the Gilliam's score 1 because Gilliam et al. described a standardized method comparing morphological alterations in the GAVE syndrome with other types of gastropathies (e.g., acute gastritis and atrophic gastritis). Results of this study strongly suggest that the distinction made by the endoscopists between GAVE syndrome and severe PHT gastropathy is supported by pathological findings. Even when Gilliam's score was compared in the areas where red spots were present, which is the corpus for PHT gastropathy and the antrum for GAVE, the differences were greater (Table 3). Because blind examination and the same methods were used in both groups of patients, these findings are probably not caused by biases. These results also strongly suggest that fibrohyalinosis is an important lesion for differentiating GAVE from severe PHT gastropathy. If this feature is added to Gilliam's score, the new score, called the GAVE score, gives increased diagnostic accuracy. Because the evaluation of spindle cell proliferation and fibrohyalinosis is as accurate as these scores (Table 4), they could be sufficient in routine evaluation. However, evaluation of these two A \ ( ~ ~L Severe PHT gastropathy i s~~atermeto~ B F GAVE ~trip'b"b~( ffuse diffuse..j ~ri sp severe "losel atter.,~/r gastropathy n[y 7- morie% Figure 3. Relationships between GAVE and severe PHT gastropathy or cirrhosis. (A) According to the clinical epidemiology, it can be observed that most of the patients with GAVE have no cirrhosis, watermelon stomach is the most frequent endoscopic feature of GAVE in patients without cirrhosis, and diffuse form is the most frequent feature of GAVE in patients with cirrhosis. (B) According to the endoscopy characteristics, it can be observed that red spots are a common endoscopic feature of GAVE and severe PHT gastropathy. signs is more subjective than the evaluation of thrombi. Therefore, we recommend a careful (on several serial sections) evaluation of the presence of all four elementary lesions and, consequently, the GAVE score. Our findings confirm that the diagnosis ofpht gastropathy is difficult to make with a conventional histopathologic examination. 31 In the literature, GAVE syndrome and severe PHT gastropathy have the following other distinctive characteristics. (1) Using endoscopy, GAVE is usually located in the antrum, whereas severe PHT gastropathy is located in the fundus and the corpus. 28 (2) Severe PHT gastropathy is specific to cirrhosis, 8 and cirrhosis is present in only 30% of reported cases of GAVE. 32 (3) About one third of patients with GAVE present with ectopic location of red spots, although ectopic features have never been associated with severe PHT gastropathy 32'33 except in one report, in which no pathological examination was performed. 12 (4) Specific pathological lesions have been described in patients with GAVE, x 'H whereas lesions in severe PHT gastropathy are not considered specific. 3~'34 (5) Finally, endoscopic patterns are different in that GAVE red spots consist of the pooling of several microred spots or red wale markings that look like microvessels, and severe PHT gastropathy lesions are the result of a diffuse erythema cut up in spots by the superposition of the mosaic pattern. 3~ These differences were recently well demonstrated using microendoscopy. 36 The relationships between GAVE and PHT gastropathy are shown in Figure 3. Whether these two features are the result of different physiopathologic mechanisms cannot be established at present. In this respect, it should be noted that the GAVE syndrome was observed in patients with more advanced liver disease than in patients with PHT gastropathy. A common mechanism seems unlikely because,

6 Januaw 1995 PORTAL HYPERTENSIVE GASTROPATHY AND ANTRAL VASCULAR ECTASIA 143 in general, GAVE and severe PHT gastropathy do not occur simultaneously in patients with cirrhosis. 33 At present, different factors are suspected in the pathogenesis of these two syndromes. Hemodynamic disturbances are suggested in PHT gastropathy because gastric perfusion alterations parallel the severity of PHT gastropathy, 37 and endoscopic sclerotherapy is a predictive factor of this syndrome. 3 Both nitric oxide and prostaglandins may be involved in the gastric mucosal hyperemia. 38 On the other hand, many investigators have suspected motility disturbances in patients with GAVE. We have recently shown that antral motility was significantly altered in patients with cirrhosis and GAVE. 39 It should be pointed out that the GAVE syndrome has distinctive characteristics in patients with and without cirrhosis. 33'4 Noncirrhotic patients with GAVE are typically older women with gastrointestinal bleeding caused by watermelon stomach and associated autoimmune connective tissue disorders. 4 However, we suspect that the clinical spectrum of GAVE syndrome in noncirrhotic patients is biased because most of these patients are selected by the presence of anemia, which is the indication at the time of endoscopy. In patients with cirrhosis, we have shown that watermelon stomach was not the predominant pattern. 33 The relationships between GAVE syndrome and cirrhosis are shown in Figure 3. It seems to us that, in fact, GAVE is not a suitable term because ectasia are present in the antrum in 64% of our patients with PHT gastropathy and because "GAVE" features have been described in the proximal part of the stomach 4 and in the duodenum or jejunum. 4. Thus, the following acronym might be proposed to better define this syndrome: GIVETHA for gastrointestinal vascular ectasia with thrombi, fibromuscular hyperplasia or fibrohyalinosis predominant in the antrum. In conclusion, the results of this study suggest that GAVE and severe PHT gastropathy are two distinct features. This conclusion must be confirmed by other studies, particularly those dealing with pathophysiology. In the meantime, these results do not support previous studies that associate PHT gastropathy with anemia, especially when red spots are found in the antrum 5 or no biopsy specimens are available. 42 References 1. Viggiano TR, Gostout CJ. Portal hypertensive intestinal vasculopathy: a review of the clinical, endoscopic, and histopathologic features. Am J Gastroenterol 1992;87: Sarfeh J, Tarnawski A. Gastric mucosal vasculopathy in portal hypertension. Gastroenterology 1987;93: Vigneri S, Termini R, Piraino A, Scialabba A, Pisciotta G, Fontana N. The stomach in liver cirrhosis. Endoscopic, morphological, and clinical correlations. Gastroenterology 1991; 101: Quintero E, Piqu6 JM, Bombi JA, Bordas JM, Sentis J, Elena M, Bosch J, Rodes J. Gastric mucosal vascular ectasia causing bleeding in cirrhosis. A distinct entity associated with hypergastrinemia and low serum levels of pepsinogen I. Gastroenterology 1987; 93: D'Amico G, Montalbano L, Traina M, Pisa R, Menozzi M, Spano C, Pagliaro L, Liver Study Group of V. Cervello Hospital. Natural history of congestive gastropathy in cirrhosis. Gastroenterology 1990; 99: Dave P, Romeu J, Messer J. Upper gastrointestinal bleeding in the patients with portal hypertension: a reappraisal. J Clin Gastroenterol 1983;5: Naparstek Y, Rachmilewitz D. UGI bleeding in a non alcoholic population with portal hypertension. J Clin Gastroenterol 1980; 2: McCormack 17, Sims J, Eyre-Brook I, Kennedy H, Goepel J, Johnson AG, Triger DR. Gastric lesions in portal hypertension: inflammatory gastritis or congestive gastropathy? Gut 1985;26: Jabbari M, Cherry R, Lough JO, Daly DS, Kinnear DG, Goresky A. Gastric antral vascular ectasia: the watermelon stomach. Gastroenterology 1984;87: Gilliam JH, Geisinger KR, Wu WC, Weidner N, Richter JE. Endoscopic biopsy is diagnostic in gastric antral vascular ectasia. The "watermelon stomach." Dig Dis Sci 1989;34: Suit PF, Petras RE, Bauer TW, Petrini PD, Petrini J. Gastric antral vascular ectasia, a histologic and morphometric study of the "watermelon stomach." Am J Surg Pathol 1987;11: Thiruvengadam R, Gostout CJ. Congestive gastroenteropathy-- an extension of nonvariceal upper gastrointestinal bleeding in portal hypertension. Gastrointest Endosc 1989;35: Sauerbruch T, Kleber G. Upper gastrointestinal endoscopy in patients with portal hypertension. Endoscopy 1992; 24: Lee FI, Costello F, Flanagan N, Vasudev KS. Diffuse antral vascular ectasia. Gastrointest Endosc 1984;30: Pugh RNA, Murray-Lyon IM, Dawson N, Pietroni MC, Williams R. Transection of the oesophagus for bleeding esophageal varices. Br J Surg 1973;60: Kervran A, Blache P, Bataille D. Distribution of oxyntomodulin and glucagon in the gastrointestinal tract and the plasma of the rat. Endocrinology 1987; 121: Chayvialle JA, Miyata M, Rayford PL, Thompson JC. Effects of test meal, intragastric nutrients, and intraduodenal bile on plasma concentrations of immuno reactive somatostatin and vasoactive intestinal peptide in dogs. Gastroenterology 1980; 79: B6rsch G, Sch~fer K, Schmidt G, Menner R, Tiedjen KU. Ectasie vasculaire antrale diffuse: cause rare d'h6morragie gastrointestinale s~v~re. Acta Endosc 1985;3: Rider JA, Klotz AP, Kirsner JB. Gastritis with veno capillary ectasia as a source of massive gastric haemorrhage. Gastroenterology 1953;24: Granai F, Smart HL, Triger DR. The relation of congestive gastropathy to anemia in portal hypertension (abstr). J Hepatol 1991; 13(suppl 2):$ Bretagne JF, Guyader D, Darnault P, Raoul JL, Siproudhis L, de Mal~zieu N, Gosselin M. Relationship between gastric mucosal changes and hemodynamic patterns in alcoholic cirrhosis. A prospective study. Gastroenterol Clin Biol 1993;17: Cales P, Cohen J, Vinel JP, Caucanas JP, Pascal JP. Existe-t-il une relation entre les signes endoscopiques gastro-oesophagiens et la pr@sence d'une an~mie chez le malade cirrhotique? Gastroenterol Clin Biol 1990;14: Lain SK. Hypergastrinaemia in cirrhosis of liver. Gut 1976;17: AIbillos A, Abreu L, Alvarez-Mon M, Gea F, Gonzalo MA, Rossi I, Garrido A, Barrios C, Escartin P. Study of the secretion of pepsinogen I in cirrhotic humans with and without portacaval shunt. Am J Gastroenterol 1988;83:37-41.

7 144 PAYEN ET AL. GASTROENTEROLOGY Vol. 108, No. I 25. Pointner H. Normal serum gastrin levels in patients with liver cirrhosis. Digestion 1975;13: Lenz H J, Struck T, Greten H, Koss MA, Eysselein VE, Walsh JH, Isenberg Jl. Increased sensitivity of gastric acid secretion to gastrin in cirrhotic patients with portacaval shunt. J Clin Invest 1987; 79: Saperas E, Ayuso RMP, Poca E, Bordas JM, Gaya J, Pique JM. Increased gastric PGE2 biosynthesis in cirrhotic patients with gastric vascular ectasia. Am J Gastroentero11990;85: Cales P, Zabotto B, Meskens C, Caucanas JP, Vlnel JP, Desmorat H, Fermanian J, Pascal JP, Gastroesophageal endoscopic features in cirrhosis: observer variability, interassociations and relationship to hepatic dysfunction. Gastroenterology 1990;98: Iwao T, Toyonaga A, Sumino M, Takagi K, Oho K, Nishizono M, Ohkubo K, Inoue R, Sasaki E, Tanikawa K. Portal hypertensive gastropathy in patients with cirrhosis. Gastroenterology 1992; 102: Tanoue K, Hashizume M, Wada H, Ohta M, Kitano S, Sugimachi K. Effects of endoscopic injection sclerotherapy on portal hypertensive gastropathy: a prospective study. Gastrointest Endosc 1992; 38: Corbishley CM, Saverymuttv S, Maxwell JD. Use of endoscopic biopsy for diagnosing congestive gastropathy. J Clin Pathol 1988; 41: Payen JL, Cales P. Modifications gastriques au cours de la cirrhose. Gastroenterol Clin Biol 1991;15: Cales P, Payen JL, Berg P, Desmorat H, Vinel JP, Pascal JP. Antral vascular ectasia: new endoscopic and clinical spectrum (abstr). Gastroenterology 1991;100:A Sallebert S, Mancheron H, Sevestre H, Dupas JL, Capron JP. Etude histomorphometrique de la muqueuse gastrique au cours de I'hypertension portale. Gastroenterol Clin Biol 1988;12: Trigger DR. Portal hypertensive gastropathy. In: Okuda K, Benha- mou JP, eds. Portal hypertension: clinical and physiological aspects. Tokyo: Springer-Verlag, 1991: Cales P, Delmotte JS, Oberti F, Fouchard I, Burtin P, Person B, Boyer J. Etude de la muqueuse gastrique du malade cirrhotique en microendoscol~ie (abstr). Gastroenterol Clin Biol 1992;16: A Panes J, Bordas JM, Pique JM, Bosch J, Garcia-Pagan JC, Feu F, Casadevall M, Teres J, Rodes J. Increased gastric mucosal perfusion in cirrhotic patients with portal hypertensive gastropathy. Gastroenterology 1992;103: Casadevall M, Panes J, Pique JM, Marroni N, Bosch J, Whittle B JR. Involvement of nitric oxide and prostaglandins in gastric mucosal hyperemia of portal-hypertensive anesthetized rats. Hepatology 1993; 18: Petit R, Charneau J, Cal~s P, Boyer J. Antral motility is altered in cirrhotic patients with or without antral vascular ectasia (abstr). Gastroenterology 1993; 104:A Gostout C J, viggiano TR, Ahlquist DA, Wang KK, Larson V, Balm R. The clinical and endoscopic spectrum of the watermelon stomach. J Clin Gastroenterol 1992; 15: Cales P, Voigt J J, Payen JL, Bloom E, Berg P, Vinel JP, Prad~re 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: Perez-Ayuso RM, Pique JM, Bosch J, Panes J, Gonzalez A, P~rez R, Rigau J, Quintero E, Valderrama R, Viver J, Esteban R, Rodrigo L, Bordas JM, Rodes J. Propranolol in prevention of recurrent bleeding from severe portal hypertensive gastropathy in cirrhosis. A multicenter randomized controlled trial. Lancet 1991;337: Received November 29, Accepted September 21, Address requests for reprints to: Paul Cal~s, M.D., Service d'hepato-gastroent~rologie, Centre Hospitalier Universitaire, F Angers Cedex 01, France. Fax: (33)