Gut 1996; 38: 759-764

Role of paraoesophageal collaterals andperforating veins on outcome of endoscopicsclerotherapy for oesophageal varices: an

endosonographic study

R K Dhiman, G Choudhuri, V A Saraswat, D K Agarwal, S R Naik

Department ofGastroenterology,Sanjay GandhiPostgraduate InstituteofMedical Sciences,Lucknow, IndiaR K DhimanG ChoudhuriV A SaraswatD K AgarwalS R Naik

Correspondence to:Dr R K Dhiman,Department of Hepatology,Postgraduate Institute ofMedical Education andResearch, Chandigarh 160012, India.

Accepted for publication20 November 1995

AbstractBackground-Endoscopic sclerotherapy(EST) is an established method forcontrolling and preventing bleeding fromoesophageal varices. However, oeso-phageal varices sclerose easily andrequire less sessions of EST in somepatients while few fail to respond. Thisstudy therefore looked at changes in thevascular anatomy of the lower oesoph-agus and upper stomach that accompanysuccessful sclerotherapy of oesophagealvarices.Methods-Endoscopic ultrasonographywas performed in 50 patients with cirrhoticportal hypertension before (control, 20patients) and after successful obliterationofvarices with endoscopic sclerotherapy ina group ofresponders (EST-R, 20 patients)and in a group of non-responders (EST-NR, 10 patients).Results-The median number and size ofsubmucosal veins at the gastro-oesophageal junction and in the loweroesophagus were significantly less in theEST-R group compared with control andEST-NR groups (p values between<0*00001 and <0.000001). Concomitantly,the number and size of paraoesophagealcollaterals were significantly less in theEST-R group compared with the othertwo groups (p values between 0.02 and0.00007). Perforating veins were identifiedin 14 (70Gb) patients in the control group,nine (900/o) in the EST-NR group and innone in the EST-R group (p<0 001 forboth controls v EST-R and EST-R v EST-NR, and p=NS, control v EST-NR).Conclusion-Oesophageal variceal sclero-sis is associated with significant reductionin the number and size of paraoeso-phageal collaterals and disappearance ofperforating veins in the lower oesophagus.(Gut 1996; 38: 759-764)

Keywords: cirrhosis, endoscopic ultrasound,endoscopic sclerotherapy, oesophageal varices, portalhypertension.

Endoscopic sclerotherapy (EST) has becomean established method of treatment of bleedingoesophageal varices in patients with portalhypertension.1-3 It is, however, unclear whyvarices sclerose easily in some patients and notin others, and also why they recur rapidly in

some who show initial sclerosis and oblitera-tion. Factors affecting these variables havebeen studied, but are still unclear.4-7 Lin et al 7found that EST is less effective in treatingpatients with oesophageal variceal haemor-rhage who have severe paraoesophageal col-laterals. McCormack et al 8 have shown thatoesophageal varices are not always the mainpathway for portal to systemic shunting andthat most shunting occurs at a deeper level.Therefore it seems that the ultimate outcomeof EST may be influenced by the para-oesophageal collaterals and perforating veins.The vascular anatomy of the lower oesoph-

agus and upper stomach especially in patientswith portal hypertension was initially deter-mined by injection of contrast or casting agentsinto blood vessels of resected postmortemspecimens.9-12 In vivo investigations likeangiography13 and percutaneous transhepaticportography14 have had the major limitation ofbeing invasive and hence not easily repeatable.Computed tomography has been used to delin-eate venous anatomy of oesophagus and upperstomach in patients with portal hypertension,but has the disadvantage of radiation expo-sure.7 15 16 The diagnostic value of this proce-dure is still unclear. Endoscopic ultrasound(EUS) has emerged as a safe, non-invasivetechnique providing good delineation of thecross sectional anatomy of this region.17 Inpatients with portal hypertension oesophago-gastric varices and paraoesophageal collateralscan be visualised adequately by endosono-graphy,18 19 and it may be repeated severaltimes to assess changes that might occur withtreatment.

Using EUS we have determined the changesin venous anatomy in the lower oesophagusand upper stomach that occur with EST insuch patients. We have prospectively studiedthree groups of patients with severe portalhypertension: (a) controls, before initiation ofelective EST, (b) responders to EST, in whomcomplete eradication of varices was achieved,and (c) non-responders to EST, in whomadequate reduction in variceal size was notachieved after a minimum of 10 sessions.

MethodsFifty patients with cirrhosis of the liver andsevere portal hypertension were studied. Theyall had a history of bleeding from gastro-oesophageal varices and had been referred forelective EST. At presentation, all patients had

759

on July 14, 2022 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.5.759 on 1 May 1996. D

ownloaded from

Dhiman, Choudhuri, Saraswat, Agarwal, Naik

TABLE I Clinical details of the patients with portal hypertension

Controls (n=20) EST-R (n=20) EST-NR (n= 10)

Age (y) 40 (30-55)* 36-5 (30-50) 42-5 (35-55.5)Sex (M:F) 15:5 14:6 7:3Aetiology of the cirrhosis

Alcoholic 4 5 2HBsAg+ 5 5 3Non-alcoholic, non-HBsAg (+) 11 10 5

Child-Pugh's classA 10 11 6B 7 7 3C 3 2 1

*Median (interquartile range).

grade 3 or 4 oesophageal varices on endoscopicexamination.20 Three groups of patients werestudied: the control group consisted of 20patients in whom EUS was performed beforestarting the first session of EST. The secondgroup consisted of 20 patients in whom com-plete oesophageal variceal eradication hadbeen achieved with repeated EST (mediansessions 8, range 3 to 10). Obliteration ofoesophageal varices was defined either asabsence of varices or the presence of very smallthrombotic veins that did not bleed after punc-ture with EST needle. These patients wereconsidered as responders to EST (EST-R).The third group consisted of 10 patients whohad not achieved reduction in variceal size bytwo grades after 11 or more sessions of EST(median sessions 13-5, range 11 to 22) andwere considered non-responders to EST(EST-NR). Ten patients who were referred forEUS examination for benign pancreatico-biliary diseases, served as controls for normalanatomy of oesophagus and upper stomach.These patients did not have any evidence ofportal hypertension on clinical and endoscopicevaluation. Written informed consent wasobtained from all patients after explaining theaims of the study.Endoscopy was performed with an Olympus

GIF-Q10 fibrescope. Oesophageal varices weregraded according to Conn's classification.20Gastric varices were scored as present orabsent. Intravariceal injections were done ran-domly using the Olympus GIF-Q10 endoscopeand NM-1K injector. Two to 4 ml of 1%polidocanol (Aethoxysclerol) was injectedcircumferentially just above the gastro-oeso-phageal junction and at 2-5 cm intervals as theendoscope was withdrawn.

Subsequent sessions ofEST were performedat two to three weeks' intervals until eradica-tion of all oesophageal varices was achieved.EUS was performed with an Olympus EU-

M3 ultrasound fibrescope and display unit.The system provides 360° radial scan with fre-quencies of 7*5 and 12 MHz. Both frequencieswere used in the examination of the loweroesophagus and upper stomach. The patientfasted overnight. After sedating the patientwith intravenous diazepam (5-10 mg), theultrasound fibrescope was inserted with thepatient in the left lateral position and passeddown into the stomach. EUS examination ofthe stomach and adjacent structures was per-formed with a water filled balloon over theultrasound probe and after the instillation of100-300 ml of deaerated water into the

stomach. The distal end of the ultrasoundprobe was placed in the body of the stomachand examination was begun by slowly with-drawing the fibrescope. EUS evaluation wasperformed at three levels, that is (a) upperstomach, (b) at the gastro-oesophageal junc-tion, and (c) 5 cm above the gastro-oesophageal junction (lower oesophagus). Ateach site, a complete 360' view of the stomachor oesophagus was obtained by adjusting theposition of the transducer. At each of the threesites submucosal and perigastric or para-oesophageal collaterals were identified, theirnumber counted, and the largest diametermeasured. The collaterals identified at or justabove the gastro-oesophageal junction weretraced approximately 5 cm in the cephaladdirection. The presence or absence of perforat-ing veins connecting the submucosal with theparaoesophageal collaterals were recorded inthe lower oesophagus. An independent assess-ment of EUS findings was also made by asecond observer.

Statistical analysisThe results are given as median and interquar-tile range. Differences between the groupswere analysed by non-parametric tests - that is,Mann-Whitney U test and Fisher's exact test.

ResultsTable I shows clinical details of patients invarious groups. These groups were age and sexmatched and no significant difference was

Figure 1: A 360° endographic scan (A) and schematicdrawing (B) of a normal lower oesophagus in a healthysubject. Note three echogenic layers alternating with twoecholucent layers. These layers correspond to (1) superficialmucosa, (2) deep mucosa, (3) submucosa, (4) muscularispropria, and (5) paraoesophagealfat tissue. A: aorta.

760

on July 14, 2022 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.5.759 on 1 May 1996. D

ownloaded from

Endoscopic sclerotherapy for oesophageal varices

Figure 2: A 180° endosonographic scan (A) and schematicdrawing (B) ofa normal upper stomach in a healthysubject. Note three echogenic layers alternating with twoecholucent layers. These layers correspond to (1) superficialmucosa, (2) deep mucosa, (3) submucosa (4) muscularispropria, and (5) serosa andfat tissue.

present in relation to aetiology of cirrhosis andChild-Pugh's class.

EndoscopyThirteen patients had grade 4 and seven

patients had grade 3 oesophageal varices in thecontrol group. No variceal column was seen inany patient in EST-R group; however, smallmucosal tags were seen in seven patients. Inthe EST-NR group, three patients had grade 4,six patients had grade 3, and one patient hadgrade 2 oesophageal varices. Gastric variceswere seen in 12 patients in the control groupand in seven patients each in the EST-R andEST-NR groups.

Figure 3: A 36U' endosonographic scan of the gastro-oesophagealjunction in a patient with cirrhotic portalhypertension before EST. Oesophageal varices (submucosalveins arrowheads) are seen within the oesophageal wall andparaoesophageal collaterals (arrows) are clearly seen

outside the oesophageal wall.

U

Figure 4: A 360° endosonographic scan of upper stomachin a patient with cirrhotic portal hypertension. Gastricvarices (submucosal veins, arrowheads) are seen within thegastric wall and perigastric collateral vein (arrow) is seenoutside the gastric wall.

Endoscopic ultrasoundOesophageal and gastric wall appeared as a fivelayer structure; three echogenic alternatingwith two echolucent layers (Figs 1 and 2).These layers have been shown to correspond tothe anatomical layers of the oesophagus andstomach as follows: the first hyperechoic layercorresponds to superficial mucosa, the secondhypoechoic layer to deep mucosa, the thirdhyperechoic layer to submucosa plus theacoustical interface between the submucosaand muscularis propria, the fourth hypoechoiclayer to muscularis propria minus the acousti-cal interface between the submucosa andmuscularis propria, and the fifth hyperechoiclayer to paraoesophageal fat or to serosa andperiserosal fat.2' Oesophageal or gastric varicesappeared as rounded or oval echo freestructures in the submucosa (Figs 3 and 4).They also appeared as longitudinal echo freestructures in the submucosa depending uponthe relative position of the varix and ultrasoundtransducer. Similarly, paraoesophageal or peri-gastric collaterals were also seen as rounded,oval or longitudinal echo free structures welloutside the oesophageal or gastric wall (Figs 3and 4). Both submucosal and paraoesophagealcollaterals appeared tortuous all along theirextent. Perforating veins were seen as anechoiccommunicating channels between submucosaland paraoesophageal collaterals (Fig 5).

Table II and Fig 6 and 7 show the effect ofEST on the submucosal veins. The number aswell as the size of submucosal veins weresignificantly lower in patients in EST-R groupthan in controls and EST-NR groups at alllevels of examination. One to two submucosalveins of 2 to 3 mm size were seen in only sixpatients in the EST-R group. These patientshad either grade 1 oesophageal varices or twoto three small mucosal tags in the loweroesophagus after the successful eradication ofvarices. In the EST-NR group, however, thenumber and size of submucosal veins were notdifferent from those in the control group.

Table III and Figs 6 and 7 show the effect ofEST on the perigastric and paraoesophagealcollaterals. The number and size of paraoeso-phageal collaterals at the gastro-oesophagealjunction and in the lower oesophagus weresignificantly less in patients in the EST-Rgroup compared with the control and EST-NR

761

on July 14, 2022 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.5.759 on 1 May 1996. D

ownloaded from

Dhiman, Choudhuri, Saraswat, Agarwal, Naik

Figure 5: A 360° endosonographic scan (A) and schematicdrawing (B) of the lower oesophagus in a patient withcirrhotic portal hypertension. A perforating vein (arrow)communicating the oesophageal varices (submucosal veins,small arrow) to the paraoesophageal collaterals (largearrow) is imaged. A: aorta.

groups. In the upper stomach while thenumber of these veins was significantly lowerin patients in the EST-R group, no significantdifference in size could be seen between thesethree groups. These findings suggest that thereis a pronounced reduction in the number andthe size of paraoesophageal collaterals and adecrease in the number of perigastric col-laterals after variceal eradication with EST.

Perforating veins were seen in 14 patients inthe control group, nine in the EST-NR group,and in none in the EST-R group (p<0O001,control v EST-R; p<0.001, EST-R v EST-NRand p=NS, control v EST-NR; Fisher's exacttest).

DiscussionWe could adequately visualise the venousmorphology of the lower oesophagus andupper stomach in all the patients. Our findingsclearly show that successful eradication of

TABLE II Number and size (median (interquartile range)) ofsubmucosal veins incontrols, EST-responders, and in EST-non-responders

p Value

Control v Control v EST-R vControl EST-R EST-NR EST-R EST-NR EST-NR

Upper stomachNumber 2 (1-3) 0 (0-1) 2 (2-4) 0.001 NS 0.0002Size (mm) 5 (3-6) 0 (0-5) 5 (4-6) 0o02 NS 0.01

Gastro-oesophageal junctionNumber 3.5 (3-4) 0 (0-1) 4 (3-4) <0 000001 NS 0.000006Size (mm) 4.75 (4-5 5) 0 (0-2) 4 (3.5-5) <0000001 NS 0 000009

Lower oesophagusNumber 3 (2-4) 0 (0-0) 3 (2-4) <0o000001 NS 0000003Size (mm) 4 (3-45) 0 (0-0) 4 (3-4) <0-00001 NS 0-000005

Figure 6: A 360° endosonographic scan of the loweroesophagus in a responder to EST (EST-R). Nosubmucosal vein (oesophageal varix) is seen within theoesophageal wall and only two to three smallparaoesophageal collaterals (arrowheads) are seen welloutside the oesophageal wall. A: aorta.

oesophageal varices by EST is associated withdisappearance of perforating veins in the loweroesophagus and reduction in the number andsize ofparaoesophageal collaterals. In contrast,there is a persistence of these vascular channelsin patients whose varices do not respond toEST.

Obliteration of submucosal veins, maydepend on early or simultaneous obliterationof the perforating and paraoesophageal col-laterals, which might be serving as feeders ofthe varices. One study showed that the degreeof embolisation (extending into feeding vessels)during EST inversely correlated with longtermrecurrence of varices, implying that obliterationof feeders might be important for effectivevariceal sclerosis.4 Lin et al7 have shown thatpatients with bleeding oesophageal varices whohave large paraoesophageal collaterals needmore sessions of EST, more sclerosant, andlonger periods to achieve variceal obliterationand have higher recurrence and repeat bleedingrates after EST. They considered that para-oesophageal collaterals and perforating veinsaffect the efficacy of EST if high pressure andblood flow continue to feed intrinsicoesophageal varices (submucosal veins) duringthe course of EST. The authors, however, didnot perform the follow up computed tomo-graphy to evaluate the effect of EST on para-oesophageal collaterals - that is, whether there

Figure 7: A 3600 endosonographic scan of the loweroesophagus in a non-responder to EST (EST-NR). Notepersistence ofsubmucosal veins (oesophageal varices,arrowheads) and large paraoesophageal collaterals(arrows). A: aorta.

762

on July 14, 2022 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.5.759 on 1 May 1996. D

ownloaded from

Endoscopic sclerotherapy for oesophageal varices

TABLE iII Number and size (median (interquartile range)) ofperigastric* andparaoesophagealt collaterals in controls, EST-responders, and EST non-responders

p Value

Control v Control v EST-R vControl EST-R EST-NR EST-R EST-NR EST-NR

Upper stomach*Number 4 (3-5) 2 (1-3) 4 (2-4) 0.001 NS 0-02Size (mm) 7-5 (5-8.75) 5 (4-6.25) 6 (5-7) NS NS NS

Gastro-oesophageal junctiontNumber 4-5 (4-6.5) 2-5 (1-3) 5 (3-6) 0 00007 NS 0.002Size (mm) 6 (5 25-7) 4 (3-5) 5-75 (5-6) 0-0002 NS- 0.01

Lower oesophagustNumber 4 (3-5.5) 2-5 (1-5-3) 3-5 (3-5) 0.0004 NS 0-02Size (mm) 5 (4-6) 3 (3-3.75) 4-75 (4-5) 0.0003 NS 0.001

is reduction in number and size of para-

oesophageal collaterals or not, after the suc-

cessful obliteration of oesophageal varices.Using percutaneous transhepatic portographyin patients with portal hypertension undergoingEST, Soderlund et a114 delineated two partsof collaterals, namely, perioesophageal andparaoesophageal varices around the loweroesophagus. Perioesophageal varices were situ-ated in the connective tissue surrounding theoesophagus, whereas paraoesophageal variceswere mediastinal veins running longitudinallydistant to or overlaying the oesophagus, or

both. The authors found that large mediastinalcollaterals were largely unaffected after success-

ful EST. It has been suggested, however, thatperioesophageal and paraoesophageal varicesmight have indicated the various degrees in sizeand extent of collaterals around the oesophagusand both correspond to the dilated adventitialveins described by Hashizume et al.7 12Percutaneous transhepatic portography in cir-rhotic patients is technically demanding andcannot be recommended as a standard clinicalprocedure; in this study important complica-tions were seen in 23% of patients with one

death caused by bleeding from the hepaticpuncture site.'4

Perhaps the greatest proportion of portosys-temic shunting of blood around the gastro-oesophageal region occurs through theperigastric and paraoesophageal collaterals,with the perforators playing a. major part inmaintaining variceal patency by connecting thetwo venous channels.8 22 Using Doppler ultra-sonography McCormack et al8 have shownthat blood flows in opposite directions at dif-ferent levels in the same varix. They noted thatduring inspiration blood flow was in a caudaddirection at 35 cm from the incisors and in a

cephalad direction at 37 cm. The direction offlow at both positions reversed at the end ofexpiration. These findings lent evidence to thepresence of a communicating vein joining thevarix between the two sites. Injection radio-graphy showed their presence and showed thatoesophageal varices were not always the mainpathway for the portal to systemic shuntingand that most shunting occurred at a deeperlevel.22 The oesophageal varices may beconsidered backwaters that arise from thetransmission of blood from the large para-oesophageal collaterals to the submucosalvarices via the perforating veins. It seems thatsuccessful treatment of oesophageal variceswith EST may depend on early obliteration of

the perforating veins. Failure of EST may beassociated with the persistence of perforatingveins as seen in our patients in the EST-NRgroup.The direction of blood flow in the

oesophageal varices may explain the partialobliteration of paraoesophageal collaterals(decrease in number and size of para-oesophageal collaterals). Hence, direct actionof sclerosant carried caudaly into the gastricvarices and peripherally into the para-oesophageal collaterals through perforatingveins might have been responsible for thechanges seen by us. Another explanation mightbe in the retrograde extension of thrombosisfrom the oesophageal to the gastric varices andparaoesophageal collaterals.23A small number of patients who failed to

respond to a large number of EST sessionswere of special interest. They had received amedian of 13.5 sessions compared with amedian of 7.5 sessions required in our hospitalexperience of over 400 cases to sclerose varices(unpublished data). Further other knownfactors that contribute to EST failure likeextravascular injections,5 associated portalvein thrombosis, and hepatocellular carci-noma,24 did not seem to be the underlyingcauses in them. None of these patients hadhepatocellular carcinoma or portal veinthrombosis and the technique of EST was thesame as for patients in the EST-R group.Therefore, we believe that failure to eradicatevarices in the EST-NR group may be relatedto the large size of paraoesophageal veinsand predominant direction of blood flow inperforating veins.Why perforating and paraoesophageal veins

get obliterated in some patients and not inothers, is not clear from our study. This may berelated to the large size of these veins and pre-dominant direction of blood flow in perforatingveins. This is a one point study in which the sizeof paraoesophageal veins were not measuredbefore starting the EST; serial assessment ofoesophageal and paraoesophageal venous cir-culation with both EUS and Doppler ultra-sound may be helpful in resolving this issue.Rapid obliteration and low recurrence rates ofoesophageal varices are also expected ifembolisation of feeders with sclerosant can beachieved by preventing sclerosant flow out ofthe varix into the systemic circulation by aballoon tamponade of the varix inserted proxi-mally during injections.4 23 A similar approachmay be useful in diverting the sclerosant flow toparaoesophageal collaterals through perforatingveins in the lower oesophagus.

In conclusion, EUS seems to be a reliableprocedure in delineating changes in the venousanatomy around the gastro-oesophageal junc-tion in patients with portal hypertension.Successful EST is associated with reduction in

number and size of paraoesophageal collateralsand disappearance of perforating veins. Thepresence of large paraoesophageal collateralsand perforating veins after an adequatenumber of EST sessions may necessitate theselection of an alternative therapy in the treat-ment of variceal haemorrhage.

763

on July 14, 2022 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.5.759 on 1 May 1996. D

ownloaded from

764 Dhiman, Choudhuri, Saraswat, Agarwal, Naik

The authors thank Anil Kumar for the preparation of the illus-trations.

1 Paquet KJ, Seussner H. Endoscopic sclerosis andesophageal balloon tamponade in acute hemorrhage fromesophagogastric varices: a prospective randomized trial.Hepatology 1985; 5: 80-3.

2 The Copenhagen esophageal varices sclerotherapy project.Sclerotherapy after first variceal hemorrhage in cirrhosis.NEnglJrMed 1984; 311: 1594-600.

3 Korula J, Balart LA, Radvan G, Zweiben BE, Larson AW,Kao HW, et al. A prospective, randomised controlled trialof chronic esophageal variceal sclerotherapy. Hepatology1985; 5: 584-9.

4 Takase Y, Shibuya S, Chikamori F, Orli K, Iwasaki Y.Recurrence factors studied by percutaneous transhepaticportography before and after endoscopic sclerotherapy foresophageal varices. Hepatology 1990; 11: 348-52.

5 Rose JDR, Crane MD, Smith PM. Factors affecting suc-cessful endoscopic sclerotherapy for oesophageal varices.Gut 1983; 24: 946-9.

6 Dilawari JB, Raju GS, Chawla YK. Development of largespleno-adreno-renal shunt after endoscopic sclerotherapy.Gastroenterology 1989; 97: 421-6.

7 Lin C-Y, Lin P-W, Tsai H-M, Lin X-Z, Chang T-T, ShinJ-S. Influence of paraesophageal venous collaterals on effi-cacy of endoscopic sclerotherapy for esophageal varices.Hepatology 1994; 19: 602-8.

8 McCormack TT, Smith PM, Rose JD, Johnson AG.Perforating veins and blood flow in oesophageal varices.Lancet 1983; ii: 1442-4.

9 Butler H. The veins of oesophages. Thorax 1951; 6: 276-96.10 Decarvalho CAF. Surl angio-architecture veineuse de la

zone de transition oesophagogastrique et son interpreta-tion functionelle. Acta Anat 1966; 64: 125-62.

11 Kitano S, Terblanche J, Kahn D, Bornman PC. Venousanatomy of the lower oesophagus in portal hypertension:practical implications. BrJ Surg 1986; 73: 525-31.

12 Hashizume M, Kitano S, Sugimachi K, Sueishi K. Three-dimensional view of the vascular structure of the lower

esophagus in clinical portal hypertension. Hepatology1988; 8: 1482-7.

13 Hoevels J, Lunderquist A, Tylen U, Simert G.Portosystemic collaterals in cirrhosis of liver. Acta RadiolDiagnosis 1979; 20: 865-77.

14 Soderlund C, Backman L, Erwald R, Forsgren L, Marions0, Wiechel K-L. Sclerotherapy of esophageal varices: anendoscopic and portographic study. Hepatology 1984; 4:877-84.

15 Bathazar EJ, Naidich DP, Megibow AJ, Lefleur RS. CTevaluation of esophageal varices. AJ3R 1987; 148: 131-5.

16 Clark KE, Foley WD, Lawson TL, Berland LL, MaddisonFE. CT evaluation of esophageal and upper abdominalvarices. Jf Comput Assist Tomogr 1980; 4: 510-5.

17 Rosch T. Endoscopic ultrasonography. Endoscopy 1992; 24:144-53.

18 Caletti GC, Brocchi E, Baraldini M, Ferrari A, Gibilaro M,Barbara L. Assessment of portal hypertension by endo-scopic ultrasonography. Gastrointest Endosc 1990; 36:S21-7.

19 Caletti GC, Brocchi E, Ferrari A, Fiorinos S, Barbara L.Value of endoscopic ultrasonography in the management ofportal hypertension. Endoscopy 1992; 24 (suppl 1): 342-6.

20 Conn HO. Ammonia tolerance in the diagnosis of esophagealvarices: comparison of endoscopic, radiologic and bio-chemical technique. Jf Lab Clin Med 1967; 70: 442-6.

21 Kimmey MB, Martin RW, Haggitt RC, Wang KY,Franklin DW, Silverstein FE. Histologic correlates ofgastrointestinal ultrasound images. Gastroenterology 1989;96: 433-41.

22 Doehner GA, Ruzicka FF, Rousselot LM, Hoffman G. Theportal venous system: on its pathological roentgenanatomy. Radiology 1956; 66: 206-17.

23 Hedberg SE, Fowler DL, Ryan RL. Injection sclerotherapyfor esophageal varices using ethanolamine oleate. Am JSurg 1982; 143: 426-31.

24 Lo G-H, Lin C-Y, Lai K-H, Malik U, Ng W-W, Lee F-Y, etal. Endoscopic injection sclerotherapy versus conservativetreatment for patients with unresectable hepatocellularcarcinoma and bleeding esophageal varices. GastrointestEndosc 1991; 37: 161-4.

on July 14, 2022 by guest. Protected by copyright.

http://gut.bmj.com

/G

ut: first published as 10.1136/gut.38.5.759 on 1 May 1996. D

ownloaded from