medical staf conference

Refer to: Cello JP: Gastroesophageal variceal hemorrhage-Path-ogenesis and management-Medical Staff Conference,University of California, San Francisco. West J Med130:531-539, Jun 1979

Medical Staf Conference

Gastroesophageal Variceal HemorrhagePathogenesis and Management

These discussions are selected from the weekly staff conferences in theDepartment of Medicine, University of California, San Francisco. Takenfrom transcriptions, they are prepared by Drs. David W. Martin, Jr., Asso-ciate Professor of Medicine, and James L. Naughton, Assistant Professorof Medicine, under the direction of Dr. Lloyd H. Smith, Jr., Professor ofMedicine and Chairman of the Department of Medicine. Requests for re-

prints should be sent to the Department of Medicine, University of Cali-fornia, San Francisco, CA 94143.

DR. NAUGHTON: * The topic for consideration to-day is variceal hemorrhage. We have asked Dr.John P. Cello, Chief of Clinical Gastroenterologyof San Francisco General Hospital, to discuss thetopic, with Dr. Robert Byers first presenting acase recently seen on the medical service.

DR. BYERS:t The patient is a 62-year-old whitewoman with an approximately two-year historyof hyperthyroidism. She was seen by her physicianone year before admission here for epistaxis andpalmar erythema. Hepatomegaly was noted to-gether with abnormal findings on liver functiontests. Results of a biopsy of the liver done at thattime were interpreted as mild persistent hepatitis.The patient was treated with prednisone in atapering dosage to 10 mg a day for the last fourmonths. She was admitted to a local hospital forhematemesis and melena 1½/2 weeks before ad-mission here; endoscopy 'at that time showedblood loss from varices. She was given a trans-fusion of 6 units of blood and sent home, but twodays following discharge she returned to the localhospital with orthostatic hypotension and blackstools. She was given another transfusion of 6units of blood and was transferred to the Univer-sity of California Medical Center. She had no

*James Naughton, MD, Assistant Chief, Medical Service, MoffittHospital, University of California, San Francisco.

tRobert Byers, MD, Medical Resident, University of California,San Francisco.

history of alcohol abuse. She was well developed,well nourished and in no acute distress; she wasalert and cooperative. Vital signs were normaland without orthostatic changes. Spider telangi-ectasias on the trunk and upper arms were noted,bibasilar rales were present on examination ofthe chest, and examination of the abdomenshowed the liver to have a 12-cm span in themidclavicular line extending 2 cm below the rightcostal margin. A firm, tender hepatic edge waspalpated. The spleen tip was felt. Ascites was notnoted, but palmar erythema was obvious. Allstool specimens were melenic.

Laboratory studies on admission disclosed thefollowing values: leukocytes, 11,000 per cu mm;hematocrit, 29.2 percent; total bilirubin, 1.3 mgper dl; serum alkaline phosphatase, 38 units perdl; lactic dehydrogenase, 173 units per ml; serumglutamic oxaloacetic transaminase, 50 units perml. Other laboratory values were normal and astudy was negative for hepatitis B surface antigen.A radiograph of the chest showed an elevatedright hemidiaphragm and small right pleuraleffusion.The hospital course was marked by a drop in

hematocrit to 25.7 percent and a transfusion of1,000 ml was required. Endoscopy showed largeesophageal varices without fresh blood loss. Someascites was found on abdominal sonography. Ex-amination of a biopsy specimen of the liver

THE WESTERN JOURNAL OF MEDICINE 531

GASTROESOPHAGEAL VARICEAL HEMORRHAGE

showed cirrhosis without hepatitis. The patientwas transferred at that time to the surgicalservice, where a splenorenal shunt operation wascarried out without complications. She was dis-charged and she returned home to the care of herprivate physician.

DR. CELLO: * A slightly more prosaic clinicaldescription of a patient with portal hypertensionwas recorded in the 17th century. John Brown,Royal Physician at St. Thomas Hospital in Lon-don, was the first clinician to describe the appear-ance of the liver in a patient with portal hyper-tension and gross ascites.1 In- 1685 Brown wrote"I send you here the figure of a hydropical (thatis, an ascitic) person. He was about 25 years ofage, a soldier in one of His Majesty's regimentshere in town who contracted his distemper bydrinking much water, when he could not stir fromhis duties and catching cold at night in beingupon the guard. His swellings resumed so therewas nothing more now to be thought of but aparacentesis, whereby we drew from the patientthree pints of brinish liquor and within three daysas much more; the next day he died." Brown de-scribes the appearance of the liver as follows:"The liver consisted in its concave, convex andinward parts of glands which made up the wholesubstance." However, it was Rene Laennec whofirst used the term "cirrhosis" in 1826 to describethe gross appearance of the liver in a patient withportal hypertension and ascites. He coined theword cirrhosis, from the Greek scirrhus, meaning,of yellow or tawny color.2

There are several aspects of portal hyperten-sion and variceal hemorrhage that should be con-sidered. First, I shall review our understanding ofthe pathophysiology of portal hypertension andthe development of gastroesophageal varices.Second, and quite distinct from the first (becausewe know that there can be varices in patients foryears without blood loss), I shall consider thepathogenesis of the actual torrential hemorrhagicepisode as a complication of varices. Third, someaspects of the medical and surgical managementof esophageal varices will be reviewed.

Pathophysiology of Portal HypertensionThe liver begins an intimate relationship with

portal venous blood (blood returning from thebowel) early in embryonic life when the ventral

*John P. Cello, MD, Chief, Clinical Gastroenterology, SanFrancisco General Hospital, Assistant Professor of Medicine,University of California, San Francisco.

hepatic diverticulum develops from the foregutinvading the omphalomesenteric vein. This em-bryonic vein receives blood from the yolk sac, astructure continuous with the primitive gut. Theomphalomesenteric vein of the embryo eventuallybecomes the portal vein in adults.

With the body at rest, between digestive peri-ods, total hepatic blood flow in an adult is 1,500ml per minute, or roughly 58 ml per 100 gramsof liver per minute. This represents 28 percent ofthe cardiac output and 20 percent of the totaloxygen utilization of the body at rest. Duringdigestion, arterial-portal venous oxygen differ-ences are notably increased due to increasingutilization of oxygen in the intestine, and the in-creased oxygen demand by the liver during di-gestion is supplied by increasing hepatic arterialblood supply. Of the 1,500 ml per minute of totalhepatic blood flow, 1,000 ml, or two thirds, isportal venous flow. In addition, portal venousblood supplies 50 percent to 60 percent of thetotal oxygen utilization of the liver at rest. Thisimportant portal flow is perfused at a centralportal venous pressure of 10 to 15 cm of water.The portal venous blood delivery system, there-fore, is a high volume, low resistance system.

In this high flow, low pressure system, anylesion in the portal venous delivery system mayincrease portal venous resistance and lead toportal hypertension. According to Dame Sherlockand many other investigators, one can conveni-ently classify portal hypertension on the basis ofthe site of increased portal venous resistance.'In extrahepatic portal hypertension, classicallyone is dealing with either portal venous occlusionin which splenic or main portal veins are throm-bosed or increased splenic blood flow due to anarteriovenous malformation or even to massivesplenomegaly. Intrahepatic presinusoidal portalhypertension will occur in cases of schistosomia-sis, in congenital hepatic fibrosis or with a varietyof portal zone-infiltrative processes that can not-ably increase portal venous resistance. Sinusoidalportal hypertension and postsinusoidal portalhypertension are seen in a variety of processessuch as cirrhosis and veno-occlusive disease ofthe liver.

Ordinarily the most common cause in the de-velopment of hypertension in our society is al-coholic liver disease (Table 1). Actually at avariety of stages in the development of alcoholicliver disease, one can find increasing portal ve-nous resistance. Cirrhosis,, however, is the stage

532 JUNE 1979 * 130 * 6


TABLE 1.-Causes of Portal Hypertension in 100Consecutive Patients Admitted to the University of

California Medical Center, San Francisco, andAffiliated Hospitals, for Portal-Systemic Shunts,

1974-1978

Alcoholic Liver DiseaseActive alcoholic ...... ...... 75Past alcoholic ........ ...... 10

HB.Ag- macronodular cirrhosis 7HB.Ag+ macronodular cirrhosis 2Hemochromatosis ....... ..... 2Schistosomiasis ........ ...... 2Wilson disease ........1.......Portal vein thrombosis ..... 1...

TOTAL .................. 100HB.Ag=hepatitis B surface antigen

in which there is a substantial irreversible com-ponent of portal hypertenson. The most importantcomponent of portal hypertension in the cirrhoticliver is the regenerative nodule.4 Several factorsare responsible for this.5 The nodule is suppliedalmost exclusively by the hepatic arterial system.New portal venous tributaries are not grown inthe hepatic tissue in the nodule. As the nodulegrows and enlarges, it will first compress hepaticvenous structures. Therefore, the nodule tends toproduce a postsinusoidal portal hypertension. Inaddition, the nodule by its growth and enlarge-ment will compress and distort the remaining nor-mal parenchyma perfused by the portal vein. Con-sequently, increased sinusoidal resistance to portalflow occurs. A third factor in the development ofportal hypertension in nodular regeneration isthe formation of small hepatic arterial-to-portal-venous shunts in the broad bands of scar tissuebetween nodules.

Varices One of Several ShuntsWhatever factors are responsible in portal hy-

pertension-be they presinusoidal, sinusoidal orpostsinusoidal, alcoholic liver disease, or post-necrotic cirrhosis-the most clinically notablesequela of cirrhosis, other than hepatic failure,is the development of a large number of portal-venous to systemic-venous shunts. The mostclinically significant shunts are those that occurat the gastroesophageai junction (producingvarices), the umbilicus (producing the caputmedusae) and the rectum (as large internalhemorrhoids). Other clinically insignificant butoccasionally prominent shunts occur around thediaphragm and around the spleen.The most important of these shunts are those

) .I 1

,p' **,-a

Figure 1.-Esophageal varices. Thiin-walled venouschannels are noted in the submucosal loose connec-tive tissue. Only the stratified squamous epithelium ofthe esophagus separates the portal blood from thelumen.

that develop at the gastroesophageal junction,where, through increased resistance to portalflow, a large number of potential venous chan-nels open. These channels, in fact, are simply thesubmucosal venous channels (the venous plexus)that are ordinarily found in the submucosa of thetubular gastrointestinal tract. Under the influenceof portal hypertension, portal venous flow ischannelled from a relatively high resistance regionin the liver to the relatively low resistance systemthat is adjacent to the esophagus; namely, theazygous and hemiazygous systemic-venous sys-tems. These variceal venous channels, therefore,are not newly formed, but rather massively di-lated, submucosal venous channels made up ofextremely thin-walled vessels that are precari-ously approximated to the lumen of the esophagus(Figure 1). Endoscopically, varices appear aslarge, soft, fleshy and tubular nodular masses thatrun longitudinally down the esophagus. In somepatients the varices are so prominent that theytouch each other across the lumen and may al-most completely occlude the lumen.

Varices, therefore, commonly develop in pa-tients with portal hypertension. However, weknow that there can be portal hypertension andvarices in patients for years without blood loss.Moreover, there can be hemorrhage from varicesthat is followed by a relatively quiescent periodof many months. Because variceal hemorrhage isthe most life-threatening consequence of cirrhosis,



TABLE 2.-Pathogenesis of Variceal Hemorrhage inCases of Portal Hypertension

Site-esophagus (from 20 to 40 cm from incisors); stom-ach: gastric varices high in cardia or fundus

Pathologic condition-small punctum or erosion; noesophagitis, ulcer or laceration

Factors responsible-unknown/undocumented* hypotension of lower esophageal sphincter and re-

flux esophagitis* Mallory-Weiss tears over a varix* esophageal/gastric peptic ulceration* surging increases in portal-pressure Valsalva maneu-

vers, straining, coughing, increasing ascites

what can be understood about the factors respon-sible for hemorrhagic episodes?We know from studies of patients and epidemi-

ology that the site of blood loss is invariably in thedistal esophagus or proximal stomach (see Table2); consequently, the proper term to describe thisblood loss would be gastroesophageal variceal hem-orrhage. The pathologic condition at the time ofendoscopy or postmortem examination is a punc-tum or small isolated erosion. It is uncommon tosee esophagitis, an ulcer or a laceration. Whatfactor is responsible? There are a variety oftheories that have been proposed to explain thedevelopment of the actual bleeding episodes. Isthere hypotension of the lower esophageal sphinc-ter that allows reflux of gastric acid and refluxof esophagitis overlying the varix? Is there aMallory-Weiss tear, a longitudinal linear tear inthe mucosa overlying the varix? Is it peptic ulcer?The factors responsible are not defined but someof these questions can be answered. In spite ofthe massive ascites possible, there is no de-monstrable hypotension of the lower esophagealsphincter in those patients in whom studies havebeen done shortly after an exsanguinating varicealblood loss. Before and after diuresis, the meanpressure of the lower esophageal sphincter inthese patients is normal.8 This is further supportedby our endoscopic and pathologic correlations.Esophagitis is not commonly found on endoscopyin the region of a varix from which there is bloodloss. Moreover, rarely do we encounter Mallory-Weiss lacerations in the region of variceal hemor-rhage. Peptic ulceration is also not commonlyfound. However, we know that there is no bloodloss from varices unless intravenous pressuresrise to dramatic levels, possibly associated withstraining, coughing or Valsalva maneuvers. Inaddition it appears that exsanguinations requiring

control through surgical therapy are unlikely tooccur unless the portal venous pressures surgeabove 45 to 50 cm of water.

DiagnosisGiven our understanding of the pathogenesis

of portal hypertension and variceal hemorrhage,how should these conditions be diagnosed? Isthere any way one can suspect, from a patient'spresentation, that an episode of blood loss isvariceal in nature? Usually variceal hemorrhageis sudden and voluminous with hematemetic epi-sodes unassociated with epigastric pain. On occa-sion, hemorrhage is less dramatic, such as wasseen in the case under discussion, in which severaldays of painless melena were noted. The latteraspect of variceal hemorrhage was generally notappreciated until endoscopy was carried out rou-tinely and we noted large eroded varices in pa-tients presenting with several days of melena.On physical examination, as exemplified in thiscase, portal hypertension may or may not besuggested. Patients with variceal hemorrhage mayhave no clinically detectable ascites, or othersigns of cirrhosis or portal hypertension. In addi-tion, on physical examination the finding of fea-tures suggestive of chronic liver disease and portalhypertension does not substantiate varices as thecause of hemorrhage. No laboratory test is spe-cific for cirrhosis and portal hypertension, andeven standard single-contrast upper gastrointes-tinal series may fail to detect varices radiographi-cally in 60 percent of the cases. Moreover, de-tection of gastroesophageal varices on bariumstudies does not confirm them as the site ofhemorrhage.

Endoscopy is the only practical method forconfirming varices as the site of upper gastro-intestinal hemorrhage. The ideal time for carry-ing out this examination is shortly after a patienthas arrived at the hospital and his condition hasbeen stabilized. Large blood clots that will ob-scure vision must be evacuated from the stomach.When the stomach is clean, fresh blood loss froma punctate area overlying a varix, or an adherentfresh clot on a varix, confirms the site of hemor-rhage.

Initial Medical TherapyVariceal hemorrhage is not a minor problem

in San Francisco. Every year 30 to 50 patientswith variceal blood loss are admitted to SanFrancisco General Hospital (Table 3) and 14

534 JUNE 1979 * 130 * 6


TABLE 3.-Causes of Significant Upper GastrointestinalHemorrhage, San Francisco General Hospital

(100 consecutive cases)

DocumentedLesion by Endoscopy

Esophagitis ............. 4Esophageal cancer ..... 1...Esophageal varices ....... 14Mallory-Weiss tear ....... 13

Total esophagus ...... 32Gastric ulcer ............ 20Gastritis ................ 14Gastric cancer ........... 3Pyloric channel ulcer ..... 6

Total stomach ....... 43

Duodenal ulcer .......... 22Duodenitis ............. 3

Total duodenum ....... 25

TOTAL ............. 100

percent of the episodes of significant upper gas-trointestinal hemorrhage occur in varices. Thelatter figure is verified in other series in urbanpopulatons.What should be the approach in cases in which

blood loss is thought to be from varices, particu-larly in patients with florid stigmata of chronicliver disease? Obviously resuscitation is the firststep; little can be done or should be done diag-nostically unless the patient's condition is firsthemodynamically stabilized. Certainly in manycases in which there is blood loss from varices,several units of blood are lost and prompt fluidreplacement is required. However, with the at-tempt to restore intravascular volume, manypatients who have had blood loss from varicesare overhydrated or hypertransfused. In prin-ciple this would not appear to be hazardousin these patients. However, some experimentaldata suggest that there is a unique intravascu-lar volume-portal pressure relationship in patientswith cirrhosis, particularly in those with ascitesin whom the ability to accommodate excessamounts of fluids is limited; this is manifested bysurging increases in portal pressure in patientswith volume overload.7 8 Because a patient withascites and cirrhosis is unable to accommodateextraordinary amounts of volume replacement, itis generally recommended that the central venouspressure be kept below 5 cm of water. In elderlypatients or patients with cardiopulmonary disease,the central venous pressure may not accurately

TABLE 4.-Proper Usage of Vasopressin (Pitressin)for Variceal Hemorrhage

Preparation: use only aqueous vasopressin, not oil.Route: intravenous continuous infusionno need for iniraarterial dose

Dosage: range for 70-kg man-0.1 to 0.4 units per minute.no loading doseno tapering needed

Patient location: intensive care environmentMajor side effects: depressed cardiac output,

hypertension,coronary ischemia

monitor left-heart filling pressures and may in-correctly estimate intravascular volumes. In thesecircumstances thought should be given to placinga Swan-Ganz pulmonary artery catheter.

In many cases in which hemorrhage is great,vomiting and water lavage may contribute to pul-monary aspiration. It may be necessary to placean endotracheal airway to preclude aspiration.Elective intubation of a patient may be of addi-tional benefit. Because many of these patients areagitated, uncooperative and alcoholic, diagnosticendoscopy and even placement of a balloon tam-ponade will be facilitated by airway intubationand muscular paralysis induced by pancuroniumbromide or succinylcholine chloride.

Vasopressin TherapyVasopressin therapy is effective in moderating

severe hemorrhage from the gut. The site of ac-tion of vasopressin infusion in patients with vari-ceal blood loss is the splanchnic arteriolar bed, andnot the varix. There is no vasopressin-responsivevascular smooth muscle in the portal venous sys-tem. At the level of the splanchnic arteriole, arelatively nonspecific vasoconstriction occurs withvasopressin therapy. This splanchnic oligemiainduced by vasopressin causes a notable decreasein portal venous flow and pressure. There is noeffect of vasopressin infusion on the hepaticartery. In fact, in some recent studies it is docu-mented that with the decrease in portal venousflow, the hepatic artery alone in the splanchnicbed actually vasodilates, maintaining near normaltotal hepatic blood flow.How should one administer vasopressin (Table

4) to a patient with variceal hemorrhage? Resultsof studies using rhesus monkeys, show that there isa stepwise decrease in superior mesenteric arterialblood flow, and a concomitant decrease in portalvenous pressure as one increases vasopressin in-fusion logarithmically.9 Earlier clinical experience



suggested that selective arterial infusion of vaso-pressin into the left gastric artery was necessary.The efficacy of selective intraarterial vasopressininfusions has been shown in recently publishedprimate studies employing electromagnetic flowmeters and manometers in the superior mesentericartery and portal veins. The intraarterial infusionof 5 mU of vasopressin per kg of body weightper minute results in a prompt decrease of thearterial blood flow and portal venous flow, anda profound drop in portal venous pressure. Atthis low dose-equivalent to 0.35 units per min-ute infusion of vasopressin in a man weighing 70kg (154.3 pounds)-systemic side effects couldbe shown with an elevation of central aorticblood pressure and a decrease in cardiac output.9Thus, continuous intraarterial vasopressin therapy,while effective in decreasing portal flow and pres-sure, does have undesirable systemic side effects(Figure 2A).With continuous systemic intravenous vaso-

pressin infusion at the same dose in primates (5mU per kg of body weight per minute), com-parable effects on splanchnic arterial flow andportal pressures are shown. A prompt decreasein superior mesenteric arterial blood flow, an in-crease in peripheral vascular resistance, and aprompt decrease in portal venous flow and pres-sure occur with intravenous infusions (Figure2B). Systemic side effects are comparable whetherthe dose of vasopressin is administered intra-arterially or intravenously.'0'11Our recommendation, based on these.and other

studies and on clinical experience, is that thevasopressin be administered by continuous intra-venous infusion at a dosage of approximately 5mU per kg of body weight per minute. Thus, wewould recommend administering between 0.1 and0.4 units per minute in an adult, as a continuousintravenous infusion. The onset of action is soprompt that no loading dose is necessary, and notapering is needed when administration of thedrug is discontinued. Another important clinicalpoint is that toxicity of vasopressin is definitelyrelated to systemic blood levels. The drug hasextraordinary toxicity in some patients with hy-pertension, low cardiac output and coronary is-chemia shown by ischemic changes on an elec-trocardiogram. Therefore a patient's conditionshould be closely monitored while vasopressin isbeing administered. The toxicity is relativelyunpredictable. In some elderly patients a fulleffective dose of vasopressin infusion may be

administered without any problems, while in someyounger patients significant coronary ischemiaoccurs.

Balloon TamponadeSengstaken-Blakemore and Linton tubes have

been used for years in patients with massive bloodloss from varices. So much morbidity and mor-tality has been associated with the use of balloontamponade therapy in cases of hemorrhage fromesophageal varices that, in some centers, the useof balloon tamponade has been abandoned infavor of vasopressin infusion or emergency sur-gical operation. Our experience at hospitals af-filiated with the University of California, San

INTRA-ARTERIAL(5 X 10-3 UNITS Kg-'min)

70FE p X SUPERIQR MESENTERIC ARTERY

o 3011- ii i

E10

.0 'i.ji +- i4.0

1102. REITANCI-I

I I III I I

0 2 4 6 8 10 12 15 25 35TIME (min)

INTRAVENOUS15 X 10-3 UNITS Kg-'lmi

80 r

E _- SUPERIOR MESENTERIC ARTERY

I 60 -

40140 _

E120_

X 60p i i_+ i PORTAL VEIN

4.0

4.0 r RESISTANCE

D _

2.0 - 3--_

0 2 4 6 8 10 12 15 25 35TIME (minl

Figure 2.-Vasopressin administered to five monkeysintraarterially (A) or to six monkeys intravenously (B).The infusions begin at minute 0 and continue for tenminutes. Superior mesenteric artery blood flow fall inboth situations within a few minutes of infusion. (Re-produced with permission from Freedman AR, andGastroenterology.9)

536 JUNE 1979 * 130 * 6

I


TABLE 5.-Suggested Physician Orders forMaintenance of Minnesota Balloon Tamponade

1. Chest radiograph, portable, to document subdia-phragmatic location of gastric balloon.

2. Keep gastric balloon inflated with 400 ml air; double-clamp inflation port; do not deflate.

3. Apply tension to balloon by taping insertion tube tocorner of mouth, using crossed wooden tongueblades, cushioning mouth with gauze.

4. Change tube position from one corner of the mouthto the other every four hours.

5. Esophageal suction port-intermittent suction; neverlavage via this port.

6. Gastric suction port-maintain on low intermittentsuction; may lavage and administer antacids/lactulosevia this port.

7. Esophageal balloon (if necessary)-inflate with airmonitoring pressure using aneroid manometer. Neverinflate above 40 mm of mercury. Deflate esophagealballoon for 30 minutes every six hours.

8. Keep patient without oral intake.9. Elevate head of bed 15 degrees.

10. Encourage good mouth care and frequent deepbreathing.

11. Tape scissors near bedside. If respiratory distresscaused by tube, cut entire tube assembly and re-move immediately.

Francisco, Medical Center has been positive. Thetamponade is effective in arresting torrentialhemorrhage from varices and allows hemody-namic stabilization in almost all cases, even inthose ultimately requiring emergency surgicaltherapy . These favorable results are explained byour restricting the insertion and maintenance oftamponade to patients who have endoscopicallyproven blood loss from varices, and by requiringthat the procedure be done by a senior gastro-enterology or surgical resident. The most commoncause of morbidity associated with the use ofballoon tamponade in other series is aspirationpneumonia. Several advances in the design andconstruction of newer balloons have helped makethis complication less likely. The Minnesota tube,the balloon tube now in use at the three hospitalsaffiliated with the University of California, SanFrancisco, has incorporated into the insertiontube, a separate lumen for aspirating esophagealsecretions. With Linton or Sengstaken-Blakemoreballoons inflated, occluding the gastroesophagealjunction, pharyngoesophageal secretions and bloodcould pool in the esophagus and spill over intothe trachea. The esophageal suction ports in theMinnesota tube, however, can prevent this and,therefore, we believe that it is the instrument ofchoice for maintaining balloon tamponade. Tomaintain effective airway cleansing in patients

with vigorous hemorrhage from esophageal varices,intubation is carried out before the insertion oftamponade balloons. Strict orders should be writ-ten and carried out for care of patients receivingballoon tamponade therapy (see Table 5). An-other interesting aspect of the new Minnesotaballoon has been successful tamponade of esopha-geal varices through the use of gastric balloonsalone. By inflating a gastric balloon with 400 mlof air and applying gentle traction to the mouth,the gastric fundic varices supplying blood to theesophageal varices are compressed and esopha-geal balloons almost never have to be used. Thelatter aspect of the balloon tamponade, esopha-geal balloon inflation, is the most difficult tomaintain and has been associated with esophagealulceration and perforation. In our experience, thelarge gastric balloon of the Minnesota tube hasbeen highly effective alone in arresting varicealhemorrhage.

Balloon tamponade is a temporizing procedure,but one that allows for more careful preparationof patients for surgical operation. In 85 percentof patients in whom balloon tamponade is used,hemorrhage will be controlled. However, in asmall percentage of patients, balloon tamponadeis ineffective. Following deflation of the balloon,blood loss will occur again in a high percentageof patients, necessitating reinsertion of the tube.Tamponade with esophagogastric balloons is aneffective hemostatic maneuver, and not the pri-mary therapy for hemorrhaging from varices.However, in our experience it is unusual forvarices to continue bleeding in spite of the useof vasopressin alone, or in combination with bal-loon tamponade.

Surgical Therapy for Variceal HemorrhageVariceal hemorrhage is often severe, occurs

and recurs unpredictably, and is difficult to treatquickly. Because of these factors, surgical decom-pressive procedures have been devised to reduceportal hypertension. Originally called Esk fistula,portal-systemic shunts have been developed overthe years and include a variety of innovative de-compressive procedures. The most commonlyused are end-to-side and side-to-side portacavalshunts, and splenorenal shunts. In patients withlarge cirrhotic livers, a variation of the side-to-side portacaval shunt can be constructed by usinga graft of Dacron or autogenous jugular vein.This "H-graft" or Drapanas mesocaval shunt isoften used at San Francisco General Hospital



TABLE 6.-A Partial Listing of Factors ShowingAssociation With Favorable Outcome Following

Portal-Systemic Shunts

Favorable Survival Factors*Younger patientsGood nutritional statusGood muscle massNonencephalopathic before shuntNo notable ascitesNonalcoholic liver diseaseAbstaining from alcoholNo chronic active hepatitis or alcoholic hyaline on biopsyAlbumin >3.0 grams per dlBilirubin <2.0 mg per dlProthrombin < two seconds prolongedSerum glutamic oxaloacetic transaminase <100

*Composite series from multiple sources-not universally agreedupon by all investigators.

TABLE 7.-The Child's Classification, Widely Used byClinicians in Assessing Heptatic Functional Reserve in

Patients With Cirrhosis

Status

Test A B C

Bilirubin (mg per dl) <2 2-3 >3.0Albumin(grams per dl) >3.5 3.0-3.5 <3.0Ascites .. . None Easily Pro-

controlled nouncedNeurologic status ... Alert Early encepha- Coma

lopathic state

in patients in unstable condition.'2 Clotting inthe prosthetic Dacron has been a problem inour series of patients, and this problem hasbeen noted by others. When an internal jugularvein is used, autogenous Drapanas shunts arewell tolerated without thrombotic complications.In many patients at Moffitt Hospital in SanFrancisco, retroperitoneal renosplenal or adreno-splenal shunts are carried out; these proceduresrequire considerable surgical expertise but havethe advantage of being constructed outside theascitic peritoneal cavity. Our experience withthese shunts has been excellent.What are the indications for portal-systemic

shunts? In whom should they be used? In pa-tients with documented varices, and in whomthere has never been blood loss from varices,shunts should not be used. Although there is ex-cellent evidence that variceal hemorrhage isavoided through the use of prophylactic shunts,decreased mortality from variceal hemorrhage inthese cases is offset by the operative mortality andexcess mortality from hepatocellular failure. Thusprophylactic portal-systemic shunts do not sub-stantively improve long-term survival in patientswith no blood loss from varices.'3"4

Therapeutic portal-systemic shunts surgicallycreated in patients who have had blood loss, orare having blood loss, from varices will arrestand prevent variceal hemorrhage. Recurrent vari-ceal hemorrhage is unlikely in those patients inwhom a technically adequate portal decompres-sive procedure has been carried out. However,when considering all patients with blood lossfrom varices, portal-systemic shunts will not im-prove long-term survival when compared withmedical treatment.'5 IX general, when therapeuticshunts are used, it appears that hepatic failureand operative mortality offset the improved mor-tality from hemorrhage that would be likely toaccrue with patients having undergone successfulportal decompression."6A variety of factors have been examined to

help predict outcomes of therapeutic decompres-sion procedures and to select those patients likelyto benefit from shunting. In spite of years ofhemodynamic study, no predictive value has beendocumented in selecting patients or shunts on thebasis of a variety of factors including portal bloodflow, hepatic artery flow, hepatic artery/portalflow ratio, portal pressures or "maximum per-fusion pressures.'7 A number of clinical factors,however, are associated with improved outcome(Table 6). We believe that Child's classificationremains the most important determinant of out-come of surgical procedure (Table 7). Patientswith a preoperative Child's classification of A orB have notably better chances of surviving sur-gical operation, and considerably better five-yearand ten-year survival, compared with patientswith Child's classification C.18-20 Nonetheless,even in patients with Child's classification C,portal-systemic shunts, though they may not im-prove five-year survival, will prevent recurrentvariceal hemorrhage. Our recommendations atSan Francisco General Hospital are that in casesof Child's classification A or B, in patients withdocumented significant blood loss from varices(greater than 3,000 ml of blood transfusion),therapeutic portal-systemic decompression shouldbe carried out when the patient's condition ishemodynamically stabilized. For patients withChild's classification C, in cases of acites, jaun-dice, malnourishment or encephalopathy, werecommend vasopressin therapy, blood transfu-sions and balloon tamponade to control hemor-rhage. We do not advise surgical procedures forthese patients unless all these other approaches

538 JUNE 1979 * 130 * 6


are unsuccessful in controlling blood loss orunless, with continued medical management, thepatients improve neurologically and metabolically.

Variceal hemorrhage is common in our urbanenvironment, especially in cases of chronic alco-holism. The medical management is temporizingat best; surgical therapy seems to improve long-term survival in only a few patients with goodhepatocellular function. It appears that the besttherapy is to prevent the most important causeof cirrhosis and portal hypertension in our society-alcohol abuse.

REFERENCES1. Brown J: A complete treatise of preternatural tumors. Lon-

don, Phil Tr Roy Soc, 1685, III, p 2482. Laennec RT: Traite de L'auscultation mediate, Paris, Chaude,

1826, II, p 1963. Sherlock S: Diseases of the Liver and Biliary System. Ox-

ford, England, Blackwell Scientific Publications, 19754. Popper H, Hutter F: Hepatic fibrogenesis and disturbances

of hepatic circulation. Ann NY Acad Sci 170:88-98, 19705. Rappaport AM, Noblauch MK, Black R, et al: Hepatic

micro-circulatory changes leading to portal hypertension. Ann NYAcad Sci 170:48-64, 1970

6. Van Thiel DH, Stremple JF: Lower esophageal sphincterpressure in cirrhotic men with ascites-Before and after diuresis.Gastroenterology 72:842-844, May 1977

7.' Boyer JL, Chatterjee C, Iber FL, et al: Effect of plasma-

volume expansion on portal hypertension. N Engl J Med 275:750-755, Oct 1966

8. Zimmon DS, Kessler RE: The portal pressure-blood volumerelationship in cirrhosis. Gut 15:99-101, Feb 1974

9. Freedman AR, Kerr J.C, Swan KG, et al: Primate mesentericblood flow-Effects of vasopressin and Its route of delivery. Gas-troenterology 74:875-878, May 1978

10. Millette B, Huet PM, Lavoie P, et al: Portal and systemiceffects of selective infusion of vasopressin into 'the superiormesenteric artery in cirrhotic patients. Gastroenterology 69:6-12,Jul 1975

11. Barr JW, Lakin RC, Rosch J: Similarity of arterial andintravenous vasopressin on portal and systemic hemodynamics.Gastroenterology 69:13-19, Jul 1975

12. Thompson' BW, Casali RE, Read RC, et al: Results ofinterposition "H" grafts for portal hypertension. Ann Surg 187:515-522, May 1978

13. Resnick RH, Chalmers TC, Ishihara AM, et al: A con-trolled study of the prophylactic portacaval shunt. Ann InternMed 70:675-688, Apr 1969

14. Conn HO, Lindenmuth WW, May CJ, et al: Prophylacticportacaval anastomoses. Medicine 51:2740, Jan 1972

15. Conn HO: Therapeutic portacaval anastomosis-To shuntor not to shunt. Gastroenterology 67:1065-1073, May 1974

16. Mutchnick MG, Lerner E, Conn HO: Portal-systemic en-cephalopathy and portacaval anastomosis-A prospective con-trolled investigation. Gastroenterology 67:1005-1019, May 1974

17. Reynolds TB: Promises! Promises! Hemodynamics andportal-systemic shunts. N Engl J Med 290:1483-1484, Jun 1974

18. Kanel GC, Kaplan MM, Zawacki JK, et al: Survival inpatients with post-necrotic cirrhosis and Laennec's cirrhosis under-going therapeutic portacaval shunt. Gastroenterology 73:679.683,Oct 1977

19. Malt RA, Sczcerban J, Malt RB: Risks in the therapeuticportacaval and splenorenal shunts. Ann Surg 184:279-288, Sep1976

20. Simert G, Persson T, Vang J: Factors predicting survivalafter portacaval shunt. Ann Surg 187:174-178, Feb 1978

Electrocoagulation: Making a Curable Cancer Incurable?IT REALLY IS AMAZING to see how well these patients have done by the electro-coagulation technique, because the method really destroys all our principles ofsurgical therapy for cancer. For example, if the cancer is confined within thewall of the bowel then one electrodesiccates, goes through, the wall of the bowelinto the fat and the fat protrudes, the natural boundary has been broken and, in

a few instances, the patient then has repeated attacks of bleeding or of infections.And, a curable cancer has eventually been made incurable by the procedure.

-CLAUDE E. WELCH, MD, BostonExtracted from Audio-Digest Surgery, Vol. 26, No. 4, in theAudio-Digest Foundation subscription series of tape-recordedprograms. For subscription information: 1577 East Chevy ChaseDrive, Glendale, CA 91206


medical staf conference

Documents