Postgrad Med J: first published as 10.1136/pgmj.34.394.424 on 1 August 1958. Downloaded from

424

THE SURGICAL TREATMENT OF PORTAL HYPERTENSION By A. G. RIDDELL, M.B.E., M.S., F.R.C.S. Senior Lecturer in Surgery, University of Manchester

It is just over io years since Whipple (I945), in TABLE I.-CLASSIFICATION OF PORTAL OBSTRUCTION his now classic paper, gave the impetus to the sur- Intrahepatic ...... 70 per cent. of cases gical treatment of portal hypertension, and this Cirrhosis has greatly increased understanding of the disease. Hepatic vein thrombosis

It is believed that a short review of the present Extrahepatic ...... 30 , ,. .. .. position would be of value at this time. The Portal vein thrombosis paper consists of two main parts, firstly, a review Combined Intrahepatic and Extrahepatic of the current views of the disordered physiology with vein thrombosis associated with portal hypertension and, secondly, Cirrhosis portal an attempt to outline a plan of treatment. Recent normal people the portal pressure may rise as high review articles of special interest are those of as 50 cm., if the patient strains. The main causes Child (I955) and Ravdin (I957) by copyright. The normal portal pressure is approximately of portal hypertension are shown in Table i. I5 cm. of saline. It is difficult to measure ac- The mechanism of portal hypertension is curately because of the difficulty in defining a obscure. Only about 30 per cent. of patients with suitable reference point, the most usual one taken cirrhosis develop portal hypertension, and there is the anterior surface of the second lumbar is no very clear correlation between the portal vertebra. The portal pressure may be measured pressure and the histological appearance of the in three ways; (i) direct measurement at opera- liver. It has been suggested that in cirrhosis there tion, (2) percutaneous splenic puncture, and (3) are three mechanisms which lead to this rise in hepatic vein wedge pressure. The first of these pressure; (i) obstruction to portal vein radicles methods gives the most reliable recording, the by fibrous tissue, (2) pressure on the portal vein http://pmj.bmj.com/ other two give a figure that is slightly lower than radicles by nodules of regenerating liver tissue, that obtained by the direct method. Percutaneous and (3) the presence of arterio-venous shunts splenic puncture is more generally useful than the within the liver. The mechanical factors are cer- measurement of the hepatic vein wedge pressure, tainly the most important. Whether an individual as it will give a true reading in the presence of patient develops portal hypertension or not, portal vein thrombosis and may probably depends on whether the collateral circu- also be combined lation is adequate to relieve portal congestion.

with a splenic portagram. The portal pressure may on September 29, 2021 by guest. Protected fall considerably when arterial hypotension occurs While ligation of the portal vein in the common at operation. laboratory animals is uniformly fatal, in monkey Portal hypertension has been defined as a pres- and in man ligation of the portal vein is nearly sure in the portal vein or its main tributaries of always followed by survival (Child, Milnes, more than 30 cm. of saline (Child, 1954). In a Holswade and Gore, 1950; Child, Holswade, series of 56 patients studied by Child (I955) at McClure, Gore and O'Neill, I952). Following operation, the portal pressure varied between 50 this procedure, either in the monkey or in man, and 25 cm. of saline; he further demonstrated portal hypertension does not develop, oesophageal that if the portal pressure was reduced, by means varices do not occur, and ascites does not appear. of a satisfactory shunt operation, to below 30 cm., That ligation of the portal vein is not followed by haemorrhage from oesophageal varices did not portal hypertension would, at first sight, make it recur. Taylor (1954), however, has shown that in difficult to understand the mechanism of portal hypertension in patients with an extrahepatic Based on a paper read to the Section of Surgery, Man- block. Probably, in these patients, there is not chester Medical Society, February I2, I957 only occlusion of the portal vein, but also of some August 1958 RIDDELL: The Surgical Treatment of Portal Hypertension 425 Postgrad Med J: first published as 10.1136/pgmj.34.394.424 on 1 August 1958. Downloaded from of its major tributaries, in such a way that the. TABLE 2.-THE MECHANISM OF AsCITES formation of an adequate collateral circulation is Low plasma osmotic pressure due to deficient albumen prevented and portal hypertension results. These synthesis. Faulty metabolism of hormones leads to water and salt findings indicate why it has been so difficult to retention. produce oesophageal varices in the experimental Increased lymph production by the liver. animal. Following ligation of the portal vein in Increased portal venous pressure. the monkey or in the dog, as a staged procedure, perioesophageal collateral veins will develop but and Baronofsky, 1953). At autopsy, in those oesophageal varices do not. However, oeso- patients who have died from haemorrhage and in phageal varices have been produced in the dog in whom the bleeding point can be identified, it is association with cirrhosis of the liver. It is almost common to see a shallow ulcer around the opening certain that oesophageal varices are not, in fact, into the vein. true collateral channels, but rather that they are Disorders of ammonia metabolism have come the result of distension of the numerous veins of into prominence in patients with liver disease and the oesophagus and cardia resulting from portal portal hypertension in recent years (McDermott, hypertension. Riddell and Adams, I954; Riddell, I955a and b). The two important clinical effects of portal Ammonia intoxication has been shown to be an hypertension are the haemorrhage from the oeso- important cause of death in patients immediately phageal varices and hypersplenism resulting from following severe haemorrhage from oesophageal congestive splenomegaly. No convincing evidence varices (McDermott, Wareham and Riddell, 1956), has been forthcoming to show that ascites will see Fig. i. Blood within the lumen of the alimen- occur in patients with an extrahepatic block and tary tract is broken down, by bacterial action, to without evidence of liver disease.' It can be stated, form ammonia; this is absorbed into the portal therefore, that ascites does not result directly from system, and in patients with portal hypertension, a raised portal pressure. The mechanism of ascites passes along the collaterals rather than through the is extremely complicated and portal hypertension liver and is, therefore, not immediately detoxi- plays only a minor role in its production. In ex- cated in the liver; as a result, ammonia intoxica- perimental animals, portal vein ligation will not tion occurs. by copyright. produce ascites, but it occurs after partial ligation Although a much clearer picture can be obtained of the inferior vena cava above the diaphragm. of This mechanically produced ascites in the experi- the pathological aspects of this disease than was mental animal has its clinical counterpart in man in the Budd-Chiari syndrome. It has been sug- gested that ascites in association with cirrhosis may sometimes be due to severe reduction in the out- flow tract of the liver (Madden, Lore, Gerold and Ravid, 1954). In the average patient with cir- rhosis, the presence of ascites indicates that there I-I0 http://pmj.bmj.com/ is severe liver-cell damage. The principal causes of ascites in association with liver disease are given in Table 2. Contrary to popular belief, the in- PARENC HKL % cidence of haemorrhoids is not greater in patients FAILURE K ) with portal hypertension than in the rest of the population, nor, if haemorrhoids develop, are they

more likely to bleed than in otherwise normal on September 29, 2021 by guest. Protected persons. Another problem of portal hypertension **as SOURCE OF AMMONA is the lack of understanding of the precise mech- -t r INGESTED LOOD anism which causes oesophageal varices to bleed. * I PROTEIN ' Digo UREA It has been known for a long time that injection of .i L DRUGS oesophageal varices is not followed by haemor- rhage; Allison (I95i) has stuck relatively large needles into varices to measure portal pressure, and this is not followed by .severe haemorrhage. At least one group of workers subscribe to the view that varices disturb the normal mechanism of the FIG. i.-The mechanism of ammonia intoxication. cardia and allow reflux of gastric contents, and this In patients with portal hypertension, ammonia is formed in the gut may pass around the liver, by followed by peptic ulceration of the varices way of collateral channels, to produce portasys- 'which leads to haemorrhage (Wagenknecht, Noble temic encephalopathy. 426 POSTGRADUATE MEDICAL JOURNAL August 1958 Postgrad Med J: first published as 10.1136/pgmj.34.394.424 on 1 August 1958. Downloaded from TABLE 3.-SOME OF THE METHODS THAT HAVE BEEN to pass an oesophageal tampon and arrest the USED IN THE MANAGEMENT OF OESOPHAGEAL VARICES bleeding to prevent both exsanguination and the (I). Measures directed at the control of the varices without attempting to reduce the portal hyperten- onset of coma. Should a balloon, after it has been sion. satisfactorily placed, fail to stop the bleeding, the (a) Balloon tamponage (Patton and Johnston, diagnosis will be in doubt and it may then be that '949). the patient is bleeding from a peptic ulcer. Peptic (b) Injection of sclerosing agents t-hrough an oesophagoscope (Crafoord and Frenckner, ulcers have an increased incidence in patients with ' 939). cirrhosis, and sometimes account for the haemor- (c) Transoesophageal ligation (Crile, I950). rhage instead of the varices. The type of tampon (d) Oesophagogastrectomy (Phemister and Hum- that is used is of some importance, because al- phreys, 1947). (e) Transverse section of the cardia (Tanner, though the varices are called oesophageal, in fact, 1950). they are not only oesophageal but also gastric, (2). Measures intended to lower portal hypertension by and haemorrhage may occur from the gastric side decreasing the arterial inflow to the portal circula- of the cardia rather than from the oesophagus. I tion. (a) Splenectomy would, therefore, advise the use of a tube which has (b) Splenic artery ligation (Everson and Cole, on it a relatively large balloon, which can be blown 1948). up inside the stomach, so that when traction is ap- (3). Measures intended to lower portal hypertension by plied it will control haemorrhage both from the facilitating the portal outflow. (a) Ligature of the hepatic artery (Gray et al., stomach and the oesophagus. 1951; Reinhoff, 1951). The tampon must have a wide-bored tube pass- (b) Omentopexy (Talma, I898). ing beyond the balloon into the stomach, so that (c) Portasystemic anastomosis (Rosenstein, 1912; suction can be applied and blood continuously Blakemore and Lord, 1945). aspirated from the stomach. Although an oesophageal tampon can be used as a temporary evident io years ago, it is in the clinical field that measure to control the haemorrhage, it cannot be the greatest advances have occurred. There are left in place indefinitely because of the danger of two reasons for operating on patients with portal oesophageal ulceration. After the tube has been

hypertension. The first and major one is in the in place for 24 hours, the balloon is let down and by copyright. treatment of haemorrhage from oesophagela traction released. Aspiration of the stomach is varices. The second is the management of continued to see whether haemorrhage has ceased. patients with hypersplenism secondary to portal If it has not done so then the balloon is blown up hypertension. At the present time there is no and traction is re-applied for a further 24-hour evidence that ascites can be treated in man by period: 48 hours is probably the limit for this operation, although the experimental work in method of treatment, and should by the end Qf animals shows promise. The majority of patients this time haemorrhage not be controlled, direct who bleed from oesophageal varices do so in a operation on the varices is mandatory. Probably series of episodes of acute haemorrhage, each one the best operation in these circumstances is the of which may endanger the life of the patient; transoesophageal ligation of the varices, either by http://pmj.bmj.com/ however, there is a smaller group in which the the thoracic route (Linton and Warren, I953), or patient may experience recurrent small haemor- by the transabdominal route (Welch, I956). This rhages which will produce portasystemic enceph- is only a temporary procedure and will not prevent alopathy, anaemia and ill health. The principal recurrence over a period of more than about three indications for treatment are as follows: months. It is, therefore, important to follow this (i) The management of the acute haemorrhage. operation with a definitive procedure to prevent

(2) Prevention of recurrent haemorrhage. further haemorrhage. In spite of the introduction on September 29, 2021 by guest. Protected (3) The treatment of hypersplenism. of new techniques, the results of treatment follow- It is axiomatic in medicine that when the treat- ing massive haemorrhage from the oesophageal ment of a disease is poorly understood the number varices in patients with cirrhosis are extremely of treatments which are available and have been disappointing. This has lead some authors. par- suggested is tremendous. Some of those recom- ticularly Welch, to urge immediate operative mended for the treatment of portal hypertension treatment to control the haemorrhage. Although are shown in Table 3. the use of the shunt operations has been advocated The essential features of the treatment of acute in the treatment of acute haemorrhage, I do not haemorrhage are the replacement of blood loss think these operations should be used in such cir- and the arrest of bleeding by oesophageal tam- cumstances, but should be reserved for the preven- ponage (Saengstaken and Blakemore, I950). For tion of recurrent haemorrhage. the reasons that have been previously given, in Numnerous criticisms have been levelled at the the cirrhotic group it is a matter of some urgency operations of portacaval and splenorenal anas- August I958 RIDDELL: The Surgical Treatment of Portal Hypertension 427 Postgrad Med J: first published as 10.1136/pgmj.34.394.424 on 1 August 1958. Downloaded from tomosis. Probably the most severe of these critics Percutaneous trans-splenic portography is of has been Sir Heneage Ogilvie (I953), who wrote: the utmost value in determining the type of ' Bleeding from the oesophageal varices raises the operation required. Details of the procedure are whole question of the surgery of portal hyperten- given by Walker, Middlemiss and Nanson (1953), sion. I know of no branch of surgery more and Atkinson,, Barnett, Sherlock and Steiner urgently in need of debunking. The subsequent (I955). Complications are infrequent, but it history of patients who have undergone these would seem advisable to limit this investigation to operations is on the average no better than that of those patients on whom it has been decided to those who have been untreated, and when they die operate. Du Boulay, Green and Hunt (I957) have the rnew opening is usually found to be closed.' stressed the importance of venography of both the Recent statistics show that this is not so. Milnes splenic and superior mesenteric veins at opera- Walker (I957) has had only two patients with re- tion. current haemorrhages out of 52 who have been In the majority of patients, both the portal and operated on by means of a portacaval anastomosis. splenic veins are patent, and there is a choice of Out of Linton's 78 patients, followed three or operation. In most instances a portacaval shunt more years, io have bled since operation, in eight should be performed rather than a splenorenal. of these the bleeding followed a splenorenal anas- This view is supported by Child (I955) and tomosis; which has been done on 6o patients in Walker (I957), but Linton (1956) is a protagonist this series, the remainder being portacavals for the splenorenal anastomosis, and, in his hands, (Liriton, 1956). The operative mortality for por- the operation gives excellent results. From my tasystemic anastomosis has fallen in recent years. own experience I would say that it is far harder to Child (I954) found that the operative mortality produce a satisfactory splenorenal anastomosis. It was 14 per cent. for a collected series of 362 opera- will, of course, not be possible to form a portacaval tions. Linton (1956) in a personal series of 114 anastomosis in patients who have an extrahepatic patients had an operative mortality of I2 per cent., block or who have thrombosis of the portal vein, but this has been reduced to 5 per cent. in the which is secondary to cirrhosis of the liver. In last 72 cases. Milnes Walker (I957) was able to these patients splenorenal operations must be

publish a personal series of 55 portacaval anas- carried out and in the majority it is possible to by copyright. toinoses with an operative mortality of 5 per cent. fashion a stoma which is large enough to remain It would seem, therefore, that two of the main patent. criticisms against portacaval and splenorenal shunt The most serious post-operative complication operations have at last been settled; the operative that will arise is hepatic coma. This can be mortality is low and the recurrent rate of haemor- prevented by the proper selection of patients and, rhage is small, since there can be no doubt that a also, by careful pre-operative care. Three days properly constructed anastomosis will remain before operation, the patient's protein intake patent. There is one further aspect of this prob- should be reduced to 20 g. a day, and he should be lem, however, which has not, in the past, re- started on a course of neomycin to decrease the ceived sufficient attention; this is whether shunt- bacterial flora of the gut. Both of these measures http://pmj.bmj.com/ ing of portal blood away from the liver will, in are of great importance in decreasing the post- fact, affect the survival time of the patient. There operative incidence of hepatic coma. After opera- is no body of evidence on which to base a conclu- tion, protein should be returned to the diet grad- sion. It is my opinion that the recent great im- ually, starting seven days post-operatively, and provements that have occurred in the after working up in steps to a protein intake of Ioo g. management of these patients ensure the success As stated earlier, the other main indication of the

of the operations. surgical treatment in patients with portal hyper- on September 29, 2021 by guest. Protected Any patient who has had a severe haemorrhage tension is the presence of hypersplenism. To per- from oesophageal varices, which required a blood form a splenectomy in these patients is to do them transfusion for its correction should be considered a great disservice, although it may control the tor a portacaval or splenorenal anastomosis. There hypersplenism, it will do nothing to control the must be convincing evidence that the bleeding was portal hypertension and they will eventually bleed from the oesophageal varices. The patient's liver again from their oesophageal varices. In many, function must be adequate to withstand the opera- the chance will be lost to perform a satisfactory tion. In assessing this, the criteria determined by shunt operation if the portal vein is already Linton (195i) are of great value. These are as thrombosed. It is probable, but not yet certain, follows: (a) no ascites should be present; (b) the that for a case with a moderate degree of hyper- serum albumen should be above 3 g. per cent.; splenism a portacaval shunt which relieves the and (c) the serum bilirubin should be below i mg. portal hypertension will cause regression of the per cent. hypersplenism. The alternative is a splenectomy, 428 POSTGRADUATE MEDICAL JOURNAL August IQ58 Postgrad Med J: first published as 10.1136/pgmj.34.394.424 on 1 August 1958. Downloaded from with a satisfactory ATKINSON, M., BARNETT, E., SHERLOCK, S., and and to follow this immediately STEINER, R. E. (i9SS), Quart. J7. Med., n.s., 24, 77. splenorenal anastomosis. BLAKEMORE, A. H., and LORD, J. W. (I945), Ann. Surg., x22, The most difficult problem of all still remains to 476. CHILD, C. G. (I954), 'The Hepatic Circulation and Portal be discussed, this is the problem of the post- Hypertension,' W. B. Saunders Co., Philadelphia. splenectomy bleeder. In those in whom the portal CHILD, C. G. (I955), New Engl. J. Med., 252, 837. CHILD, C. G., HOLSWADE, G. R. McCLURE, R. D., GORE, vein is still patent, a portacaval anastomosis should A. L., and O'NEILL, E. A. (1952), Surg. Gynec. Obstet., 94,31. be attempted, although in these circumstances it CHILD, C. G., MILNES, R. F., HOLSWADE, G. R., and GORE, A. L. (I950), Ann. Surg., 132, 475. will not, with certainty, cure the patient (Walker, CRAFOORD, C., and FRECKNER, P. (I939), Acta oto-laryng., 27, 1957). In some patients after splenectomy, the 422. thrombus in the splenic vein may spread back to CRILE, G. (1950), Arch. Surg. (Chicago), 6i, 654. Du BOULAY, G. H., GREEN, B., and HUNT, A. H. (957), Brit. block the entrance of the left gastric vein into the med. Y., i, I89. splenic vein. In those in whom the portal vein is EVERSON, T. C., and COLE, W. H. (1948), Arch. Surg. (Chicago), 56, 153. thrombosed the results of any form of treatment GRAY, H. K., MANN, F. C., BOLLMAN, J. L., and are depressing. Probably the best form of opera- GRINDLAY, H. J. (95I), Ann. roy. Coil. Surg. Engl., 8, 354 LINTON, R. R. (I95I). Ann. Surg., I34, 433. tion in these patients is a relatively radical proximal LINTON, R. R. (1956), 'Portal Hypertension in Disease of the gastrectomy, excising the lower part of the oeso- Liver,' Pitman, London, Edit. Schiff, L. to obliterate the varices, and removing a LINTON, R. R., and WARREN, R. (1953), Surgery, 33, 243. phagus MADDEN, J. L., LORE, J. M., GEROLD, F. P., and RAVID, liberal portion of the proximal part of the stomach J. M. (i9s4), Surg. Gynec. Obstet., 99, 385. and McDERMOTT, W. V., RIDDELL, A. G., and ADAMS, R. D. to reduce the acid secretion (Phemister (I954), Ann. Surg., 140, 539. Humphreys, I947). An alternative to this is to McDERMOTT, W. V., WAREHAM, J., and RIDDELL, A. G. carry out the operation advised by Tanner (I950), (1956), Ibid., f44, 318. stomach is transected and resutured OGILVIE, H. (I953), Surgery, 34, 768. in which the PATTON, T. B., and JOHNSTON, C. G. (I949), Arch. Surg. immediately below the cardia. (Chicago), 59, 502. PHEMISTER, D. B., and HUMPHREYS, E. M. (I947), Ann. The management of patients with portal hyper- Surg., 126, 397. tension is a fascinating problem and has produced RAVDIN, I. S. (1957), Ann. roy. Coll. Surg. Engl., 20, 7I. a very great challenge to surgeons. I have RIDDELL, A. G. (1955), Postgrad. med. J., 31, 389. attempted in this article to outline some of the RIDDELL, A. G. (I955), Ann. roy. Coll. Surg. Engl., 17, 319. W. F. RIENHOFF, (I95I), Johns Hopk. Hosp. Bull., 88, 368. by copyright. problems that have still to be answered, and have ROSENSTEIN, P. (1912), Arch. klin. Chir., 98, I082. also shown that there is now some evidence that SAENGSTAKEN, R. W., BLAKEMORE, A. H. (I950), Ann. this disease can be treated on a rational basis with Surg., 131, 78I. TALMA, S. (I898), Berlin klin. Wschr., 35, 833. surgery. I hope I have dispelled the impression TANNER, N. C. (I950), Proc. roy. Soc. Med., 43, I47. that the operations of portacaval and splenorenal TAYLOR, F. W. (I954), Ann. Surg., I40, 652. anastomosis are not without their value, indeed, WAGENKNECHT, T. W., NOBLE, J. F., and BARONOFSKY, the best we have to offer at the present I. D. (l953), Surgery, 33, 869. they are WALKER, R. M. (1957), Lancet, i, 57. time in the treatment of this disease. WALKER, R. M., MIDDLEMISS, J. H., and NANSON, E. M. (1953), Brit. 3'. Surg., 40, 392. BIBLIOGRAPHY WELCH, C. S. (I956), New Engl. 3J. Med., 255, 677.

ALLISON, P. (igsi), Thorax, 6, 325. WHIPPLE, A. 0. (I945), Ann. Surg., 122, 449. http://pmj.bmj.com/ on September 29, 2021 by guest. Protected NOTICE OF SPECIAL INTEREST TO SUBSCRIBERS: 'WHY NOT HAVE YOUR COPIES OF THIS JOURNAL BOUND INTO YEARLY VOLUMES?' HAVE You can have your twelve monthly issues fully bound in dark green pin head YOUR cloth, lettered in gilt on spine with name of Journal, Volume Number and year, complete with index at front, for 22s. 6d. post free. A limited number of out of fd l print ournals are available to bind into volumes and make your library complete. Prine on application giving details of issues required to complete back volumes. THE FELLOWSHIP OF POSTGRADUATE MEDICINE BOU 60 PORTLAND PLACE, LONDON, W.l'