Efficacy and Risks of the Distal Splenorenal Shunt in the Treatment of Bleeding Esophageal Varices Ft. Mosimann, MD, Lausanne, Switzerland P. Loup, MD, Lausanne, Switzerland
In 1967, Warren, Zeppa, and Fomon  proposed the treatment of certain cases of esophageal varicosities due to portal anastomosis,
by a distal splenorenal
with a gastric
ization.” This technic, using the short vessels, the spleen, and the splenic vein as a shunt, aims to electively decompress the esogastric varicosities, a habitual source of hemorrhage, in contrast with the
classic shunts which achieve a decompression of the entire portal system. According to Warren, Zeppa, and Fomon [l 1,this shunt is essentially designed for patients free of ascites and still possessing a satisfactory portal hepatic perfusion. Theoretically, the advantage of the shunt is to maintain irrigation of the liver by the portal blood and so prevent an accelerated deterioration of liver functions as well as the appearance of a postshunt encephalopathy. This method of surgical treatment of portal hypertension has been controversial, and despite positive results reported in the literature [2-111, the Warren shunt has as yet been little used, particularly in Europe. Presented herein are the results obtained by this type of operation in the University Hospital of Lausanne from 1969 to April 1976. Material and Methods From 1969 to April 1976, we performed a distal splenorenal anastomosis in twenty-two patients (15 males, 7 females) ranging in age from thirty-two to seventy years (average, 54 years). All the patients had massive gastrointestinal hemorrhage; twenty were stabilized at the time of operation and two underwent emergency surgery. The portal hypertension was due to alcoholic cirrhosis in twelve patients, posthepatitic cirrhosis in eight, primary biliary cirrhosis in one, and alcoholic hepatic steatosis in
From the Department of Surgery B, University Medical Center, 1011 Lausanne. Switzerland. Reprint requests should be addressed to Professor R. Mosimann, MD, Department of Surgery B. University Medical Center, 1011 Lausanne. Switzerland.
Volume 133, February 1977
one. According to Child’s classification, sixteen patients belonged to group A, five to group B, and one to group C. In twelve patients we’ performed a terminolateral splenorenal anastomosis with gastric devascularization, as proposed by Warren, Zeppa, and Fomon [I]. In the ten other patients, we achieved a terminoterminal splenorenal anastomosis with ligation of the renal vein downstream from the junction with the adrenal and gonadal veins. Also, the splenic artery was ligated in seven patients with a significant hypersplenism and the inferior mesenteric vein was ligated in eight patients. Appreciation of the results is based on the following criteria: history, clinical and biologic examinations, barium swallow, and sometimes celiac angiography; also, five patients underwent postmortem examination.
Results In an effort to evaluate the operative risk as well as the efficacy at the operation, we will study successively the immediate, intermediate, and long-term results. Immediate Results
Postoperative Complications. Postoperative complications included transient ascites (17 patients), obstruction of the shunt (2 patients; 1 died), stress ulcer (1 patient, who died), and retrogastric lymphocele (1 patient). Two of our twenty-two patients died in the immediate postoperative period, for a mortality of 9.1 per cent. The first, a thirty-three year old male, died eighteen days after surgery (terminolateral anastomosis) of hemorrhage from a duodenal stress ulcer followed by acute hepatic insufficiency. This patient had posthepatitis cirrhosis and had the biologic and clinical characteristics of Child’s group C. The second patient, a forty-three year old male, had an alcoholic cirrhosis and died on the nineteenth postoperative day. At surgery, we noticed a spleen of small size and a splenic vein with a fibrosed wall of only 8 mm diameter. There was a well developed dilated lymphatic network in the retroperitoneum. Dissection of the splenic vein and left renal vein was
Mosimann and Loup
accompanied by an abundant lymphorrhea. A terminoterminal splenorenal anastomosis was achieved under difficult conditions, with the left renal vein inhabitually situated above the splenic vein. Despite having no ascites at the time of surgery, the patient lost a large amount of ascites in the postoperative period. On the fifth postoperative day, a massive gastrointestinal hemorrhage led to reexploration which revealed thrombosis of the shunt. Splenectomy and an anastomosis between the thoracic duct and the left jugular vein were performed in the hope of stopping ascites formation. The patient went into septic shock, followed by a clear improvement in his condition and then a recurrence of the hemorrhage. A third laparotomy revealed thrombosis of the portal vein, and thrombectomy followed by laterolateral portacaval anastomosis was performed on the fourteenth postoperative day. Hemorrhage recurred again; a fourth operation showed a permeable portacaval shunt. The patient died in hepatic coma nineteen days after the first operation. Regarding postoperative morbidity, we noticed recurrence of bleeding in a thirty year old alcoholic female with cirrhosis who underwent a terminolateral portacaval anastomosis for thrombosis of the splenorenal shunt on the thirty-sixth postoperative day. This patient died one year later of hepatic insufficiency without rebleeding. Another patient, a sixty-one year old male, with alcoholic cirrhosis, had a retrogastric lymphocele requiring drainage ten days after surgery. Intermittent lymph drainage persisted during three months and then progressively ceased. The postoperative course was unremarkable in the other patients, although transient postoperative ascites was diagnosed or suspected in seventeen patients. Evolution of the Vuricosities. Barium swallow performed on seventeen patients two weeks to two months after operation showed persistence of the varicosities in four patients, regression of the varicosities in seven, and disappearance of the varicosities in six. Intermediate and Long-Term Results
Nineteen of the twenty operative survivors are included in the intermediate and long-term results, excluding one patient who underwent portacaval anastomosis after thrombosis of the distal splenorenal shunt. Mortality. Five patients have died, four of them from hepatic insufficiency without rebleeding (3 alcoholic cirrhosis and 1 primary biliary cirrhosis). The first of these four died at home in hepatic coma three months after surgery. The second patient died three 164
months after distal splenorenal anastomosis, several days after surgery for fracture of the neck of the femur. The third patient died two years after surgery and one month after orthopedic surgery. The fourth patient died three and a half years after surgery while he was hospitalized for bronchopneumonia and sepsis; postmortem examination showed multiple gastric erosions, a hepatoma, and a normal permeable shunt. One patient with alcoholism died of gastric hemorrhage after thrombosis of the portal vein six months after surgery while he was hospitalized for his seventh episode of hepatic decompensation. Recurrence of Hemorrhage. As previously noted, there were three recurrences of bleeding in the immediate postoperative period: two patients with shunt obstruction (1 death) and one patient with stress ulcer (who died). Of the nineteen other patients, for whom follow-up varies between three months and seven years, two had recurrence of hemorrhage. One had gastric hemorrhage and died six months after surgery; postmortem examination showed that hemorrhage was not due to esophageal varicosities but rather to ulcerous stasis gastritis after fresh thrombosis of the portal vein with a patent although slightly stenosed splenorenal shunt. The other patient reported intermittent vomiting of a small amount of blood over a two week period, after alcoholic abuse, without evidence of melena. The source of the hemorrhage could not be located and angiography revealed good permeability of the shunt. None of the other seventeen patients rebled. Permeability of the Shunt. As previously mentioned, we observed postoperative shunt thrombosis in two of our twenty-two patients. This complication was associated with a massive recurrence of hemorrhage. The shunt proved patent in nineteen of the other twenty patients: five by angiography, four at autopsy (at 18 days, 3 months, 6 months, and 3.5 years after surgery), and ten on barium swallow (6 with disappearance and 4 with marked regression of the varicosities). Among the four autopsies, two involved terminolateral and two terminoterminal anastomoses. In one patient, the terminoterminal anastomosis was slightly stenosed. Angiography revealed a normally permeable shunt in each of five patients showing persistence or minimal regression of varicosities on barium swallow. These five patients represented three terminolateral and two terminoterminal anastomoses. In one of these the varices disappeared several months later (Figure 1) and regressed markedly in two others one and a half years and four years after surgery. Patency is doubtful in one remaining patient with persistent varicosities on barium swallow who could The American Journal of Surgery
Warren for Shunt
Figure 1. Barium swallow showing progressive disappearance of varicostties. Left, two wee&s postoperatively. Mtddk, seven weeks postoperatively. Right, two and a half years postoperatively.
not undergo angiography but has also not rebled six months after surgery. In addition, angiography revealed two interesting facts: (1) in two patients angiography showed varicosities in the pyloroduodenal area (Figure 2) that were not present on preoperative angiography, with large folds on barium swallow (Figure 3); and (2) angiography showed an important arterial suppleance to the spleen, through the right gastroepiploic artery, in one patient who had ligation of the splenic artery (Figure 4). Liver Function. Considering nineteen of the twenty operative survivors, of the four who died of hepatic failure two died after an orthopedic operation although their liver function had been satisfactory, one died at home three months postoperatively without any further follow-up (alcoholic abuse seems to be the cause of the hepatic deterioriation), and the fourth one died of bronchopneumonia with sepsis. Volume 133, February 1977
Twelve of the fourteen currently surviving patients have been examined with regard to liver function. (Table I.) All of these patients have an abnormal electrophoresis characteristic of cirrhosis but without appreciable change between the pre- and postoperative results, except an aggravation of hypergammaglobulinemia and hypoalbuminemia in one case. Levels of direct and total bilirubin, serum glutamic oxalacetic transaminase (SGOT), serum glutamic pyruvic transaminase (SGPT), albumin, and prothrombin also show little modifications. Encephaiopathy. We had no information about the patient deceased at home from hepatic insufficiency three months after surgery. Among the other patients, we have had only one case of transient encephalopathy. This occurred in a sixty-six year old male with alcoholic cirrhosis who for several weeks after surgery presented with behavioral changes, memory difficulties, and increased inability to con165
Mosimann and Loup
Figure 2. AngbgraMy thirty-one months postoperatively showing large gastric varices and varicosities In the pyloroduodenal area (shunt patent).
he is now well and works normally without any treatment. Hypersplenism. Among twelve patients studied the platelet count is unchanged postoperatively in three patients (including 1 patient with ligated splenic artery), elevated in seven patients (including 4 with ligated splenic artery), and decreased in two patients (1 with ligated splenic artery). Other Complications. One patient developed gastric and duodenal ulcers without hemorrhagic episode eight to ten months after surgery responsive to medical treatment; healing was confirmed by endoscopic examination. Apart from transitory postoperative ascites seen in the majority of our patients, no patient had discernible subsequent ascites; three patients are restricted to a salt-free diet. No alteration in kidney function was observable in our ten patients who underwent terminoterminal splenorenal anastomosis with ligation of the left renal vein. centrate;
Figure 3. Barium swallow in the same patient as Figure in 2 showing large folds and varicosities in the duodenum.
The distal splenorenal shunt has a relatively acceptable operative mortality in the immediate postoperative period (2/22; 9.1 per cent). One of these deaths with hemorrhage from a stress ulcer and associated liver failure is not directly attributable to the type of procedure itself. The death of the second patient is related to the indications for this particular type of procedure. The presence of a narrow splenic vein with wall fibrosis, as well as the unfavorable location of this vein with respect to the left renal vein, hindered construction of a shunt under good conditions and carried with it a risk of thrombosis of the The American Journal of Surgery
Shunt for Esophageal
F&we 4. Celiac at#[email protected]
four and half months postoperatively showing sUPpleaWe to the spleen thtvugh a artery after We gastro+p&loic right ligation of the splenic artery.
shunt. With these factors in mind, we surmise that a classic type of shunt would have been more ideal than a Warren type of shunt, especially since there existed a considerable stasis in the retroperitoneal lymphatic system. Considering this experience, we believe that a splenic vein of small caliber less than 1 cm in diameter and a lymphatic stasis, under the same heading as an ascites, constitute contraindications to the distal splenorenal shunt and are indications for a portosystemic derivation to achieve hepatic decompression. The risk of thrombosis of the distal splenorenal shunt is small. We observed this complication in two of our twenty-two patients-and in the immediate postoperative period only-manifested by a recurrence of hemorrhage. Among the twenty other patients, shunt permeability was confirmed on nine occasions by autopsy or by angiography. The disappearance or the regression of the esophageal varicosities leads us to strongly suspect that the shunt is functional in ten other patients. As angiography could not be performed, some doubt persists concerning shunt patency in one patient whose varicosities have not regressed, although no bleeding has recurred. We should remark that the persistence of esophageal varicosities does riot signify automatically that the shunt is obstructed (as we have been able to show by angiography in 5 patients) and that varicosities can regress after several months. The absence of recurrent bleeding, independent of radiologic findings, appears to us to be also a valid argument in favor of shunt permeability. In reality, in thrombosis of the anastomosis the return of splenic Volume 133, February 1377
(12 patients) Postoperative
Bilirubin (mg/lOO 1.2 1.2-2 2.0 Albumin (gm/lOO 3.5 3 .o-3.5 SGOT and SGPT Normal Slightly elevated Prothrombin 70% 50-70% 50%
ml) 6 4 2
6 4 2
9 2 1
6 4 2
blood is assured only by the short vessels and the gastroesophageal veins, and one can anticipate development of an appreciable venous hypertension in this region with recurrence of hemorrhaging. Although we have the impression that terminoterminal anastomosis achieves better hemodynamic conditions than terminolateral anastomosis, we have not been able to prove a marked difference between the two variants. It is probable that the persistence or disappearance of the varicosities depends less on the type of splenorenal anastomosis than in the number and dimension of the short vessels. Therefore, the decompression of the esogastric varicosities by the distal splenorenal shunt appears effective to us. However, a patent anastomosis does not always prevent the development of varicosities in the antropyloroduodenal area as angiography 167
Mosimann and Loup
showed us in two instances with large folds demonstrated on the barium swallow. These varices could explain recurrence of bleeding after successful Warren shunting. Our patient with thrombosis of the portal vein and hemorrhagic ulcerative stasis gastritis with a patent anastomosis seems to confirm such a possibility. We have not observed an enlargement of esophageal varices with bleeding in presence of a patent shunt, as it has been reported recently . As for the influence of the distal splenorenal shunt on liver function, the number of cases is too limited and the follow-up too short to draw definite conclusions. Four deaths are caused by hepatic failure, but two followed an orthopedic operation and one a bronchopneumonia with sepsis in a patient with hepatoma; the fourth patient proceeded with alcoholic abuse. Among twelve survivors contacted and examined, only one shows a moderate deterioration of liver function. The occurrence of postoperative encephalopathy is minimal after Warren shunting. We observed this complication of moderate severity only once. It appeared soon after surgery for a few weeks only and we believe that it must be attributed to transient postoperative liver decompensation. The role of ligation of the splenic artery in hypersplenism is suspect. Among twelve patients with follow-up, the hypersplenism was as follows: not altered in three (1 of whom had splenic artery ligation); decreased in seven (4 with splenic artery ligation); and increased in two (1 with splenic artery ligation). The marked development of arterial collaterals occurring after ligation of the splenic artery that we have seen in angiography in one patient makes us doubt the utility of the ligation. Furthermore, it is difficult to evaluate the role of the arterial ligature and that of the splenic decompression in the regression of the hypersplenism. The risk of development of persistent ascites can be considered as negligible if the indication for a distal splenorenal shunt is correct, that is, if there is no ascites before the operation and if the retroperitoneal lymphatic net is moderately developed. Summary
Analysis of twenty-two patients who underwent distal splenorenal anastomosis shows that the operative mortality and morbidity are low if the indications for the procedure are strictly adhered to. The
decompression of the esophageal varicosities is effective, even in the presence of persistent varicosities. This decompression is probably less important and perhaps slower than after portacaval shunting, so we now do not recommend distal splenorenal shunting as an emergency operation. Thrombosis of the shunt is rare and related especially to local anatomic conditions and operative technic. Shunt thrombosis is an early complication that we observed on two occasions manifested by recurrence of hemorrhage. There does not appear to be an appreciable.difference between the terminoterminal and terminolateral distal splenorenal anastomoses. The effect of ligation of the splenic artery in the hypersplenism is doubtful. The frequency of postanastomotic encephalopathy is minimal with this type of shunt. Liver function tests were little changed during the period of postoperative observation, but our results do not permit us to conclude that the distal splenorenal anastomosis less unfavorably affects liver function than do classic shunts. References 1. Warren WD, Zeppa R, Fomon JJ: Selective transsplenic decompression of gastro-esophageal varices by distal splenorenal shunt. Ann Surg 166: 437, 1967. 2. Carlson RE, Ehrenfeld WK: Recurrent variceal hemorrhage following successful Warren shunt. Arch Surg 111: 567, 1976. 3. Hutson DG, Perelras R. Zeppa R, Levi JU. Schiff ER, Fink P: The fate of esophageal varices following selective distal splenorenal shunt. Ann Surg 163: 496, 1976. 4. Moreno C, Banet R, Cajigal RG: Selective splenic-renal decompression in the surgical treatment of portal hypertension. Chir Gastroenterol6: 195, 1974. 5. Reichle FA: Portal hemodynamics after distal splenorenal (Warren) shunt. Ann Surg 176: 195, 1972. 6. Salam AA, Warren WD, LePage JR, Viamonte MR. Hutson D, Zeppa R: Hemodynamic contrasts between selective and total portal-systemic decompression. Ann Surg 173: 627, 1971. 7. Silver D, Puckett CL, McNeer JF. McLeod ME, Sabiston DC Jr: Evaluation of selective transsplenic decompression of gastroesophageal varices. Am j Surg 127: 30; 1974. 6. Thomford NR. Sirinek KR, Martin EW Jr: A series of 20 successful Warren shunts..Arch Surg 110: 564, 1975. 9. Warren WD, Fomon JJ, ZeppaR: Further evaluation of selective decompression of varices by distal splenorenal shunt. Ann Surg 169: 652, 1969. 10. Warren WD, Salam AA: Surgery for the portal hypertension of cirrhosis: the need for change, chapt 14, p 127. Portal Hypertension: Major Problems in Clinical Surgery (Child CG Ill, ed). Philadelphia, WB Saunders, 1974. 11. Warren WD, Salam AA, Hutson D, Zeppa R: Selective distal splenorenal shunt. Technique and results of operation. Arch Surg 106: 306, 1974. 12. Zeppa R, Warren WD: The distal splenorenal shunt. Am J Surg 122: 300, 1971.
The American Journal of Surgery