- Email: [email protected]
Further report of a prospective randomized trial comparing distal splenorenal shunt with end-to-side portacaval shunt
GASTROENTEROLOGY
1985;88:424-9
Further Report of a Prospective Randomized Trial Comparing Distal Splenorenal Shunt With End-to-Side Portacaval Shunt An Analysis of Encephalopathy, Quality of Life
Survival, and
B. LANGER, B. R. TAYLOR, D. R. MACKENZIE, R. M. STONE, and L. BLENDIS Department of Surgery, University of Toronto: General Hospital; and Departments of Surgery, Hospital, Toronto, Ontario, Canada
We electively compared the distal splenorenal (“selective”) shunt with the end-to-side portacaval shunt in 80 prospectively randomized patients with variceal bleeding. Selective shunts required more operative time (3.9 vs. 2.8 h) and blood replacement (4.6 vs. 2.5 LJ) and postoperative mortality was slightly higher (5 of 38 selective vs. 0 of 40 portacaval). Postoperative complication rates were similar. After 65-mo mean follow-up, both shunts have protected well against late gastrointestinal bleeding (5 selective, 4 portacaval episodes). However, after selective shunts, spontaneous encephalopathy occurred less often (23% vs. 40% of patients), was severe in fewer patients (12% vs. 33%), and precipitated fewer hospital admissions (6 admissions in 4 selective patients vs. 26 admissions in 13 portacaval patients). Furthermore, selective shunt patients remained longer without functional disability (83% vs. 70% of postoperative patient months). Long-term survival was not significantly different in the two groups (5yr survival: selective 51% portacaval 56%). In 1967, Warren and colleagues concept of selective decompression
(1)
introduced the of esophagogas-
Received February 24, 1983. Accepted August 16, 1984. Address requests for reprints to: B. Langer, M.D., Toronto General Hospital, Eaton Building N, 9-236, Toronto, Canada M5G lL7. This work was supported by the General Surgery Research Fund, Toronto General Hospital. The authors thank Denise Reynolds for manuscript preparation. 0 1985 by the American Gastroenterological Association 0016.5085/85/$3.30
T. GILAS,
Departments of Surgery and Medicine, Toronto Western Hospital and Mount
Toronto Sinai
tric varices for patients with portal hypertension and bleeding. This procedure was developed to avoid complete portal blood diversion, which occurs after standard total shunt procedures such as portacaval or mesocaval shunts. This total diversion was believed to be the major contributing factor in the development of postshunt encephalopathy (PSE) and late mortality (2). The effectiveness of the selective shunt in controlling bleeding has been documented in several reports (3-5). The technical difficulty in its performance has also been noted. With experience, however, the operative morbidity and mortality can be as low as that of standard total shunts (6). A lower PSE incidence following the selective shunt has been documented in two controlled trials from other centers (7,8]. The third trial (9) failed to show any difference in encephalopathy rates between selective and total shunts. An improvement in long-term survival has been claimed in nonalcoholics based on uncontrolled data (10); however, improved survival has not yet been shown in any of the controlled trials (7-9). In 1971, we began a prospective randomized trial comparing the distal splenorenal shunt (“selective”] with the end-to-side portacaval shunt (“portacaval”). A preliminary report of this elective, therapeutic trial in 1980 (11) stated that the operative mortality in the selective shunt was higher than in the portacaval shunt, but the postshunt encephalopathy Abbreviation athy.
used
in this
paper:
PSE, postshunt
encephalop-
February
1985
rate was significantly lower. There was then no difference in long-term survival rates, although the data suggested a trend favoring the selective shunt. This report is a further follow-up of the 56 patients originally reported and an additional 24 patients who have been entered until January 1981 and evaluated until April 1982.
Methods As reported previously, all patients in the trial were biopsy-proven cirrhotics with portal hypertension who had survived at least one major variceal hemorrhage (requiring a minimum 3-U transfusion at our institutions or elsewhere). Actively bleeding patients received intensive care support under a combined medical-surgical team. Nonoperative treatment, including vasopressin and balloon tamponade, was used to control bleeding in the first 24 h. If bleeding continued or recurred, and the patient was considered operable, an emergency portacaval or mesocaval shunt was performed. If the bleeding was controlled medically, patients were stabilized and then electively considered for entry into the distal shunt trial. Jaundice, previous encephalopathy, and controlled ascites were not contraindications to entry into the trial. Endoscopy, liver biopsies, and angiography were carried out in all patients before entry to document the nature of liver disease, to confirm that there were no other upper gastrointestinal tract lesions to account for the bleeding, and to demonstrate that splenic and renal veins were suitable for shunting. Patients were excluded if they refused randomization, were over 70 yr of age, had hepatofugal portal flow, or had severe alcoholic hepatitis by histologic criteria (>lO% of cells with hyaline) or on clinical grounds (fever, leukocystosis, and tender liver). Also excluded were those with associated lesions requiring simultaneous treatment (peptic ulcer or hiatus hernia], severe uncontrollable ascites, severe coagulopathy [refractory to administration of clotting factors), or persistent spontaneous encephalopathy. If such patients later improved, they were again considered for the trial. Written informed consent was obtained and randomization was carried out in the operating room by surgeon’s coin toss (to the floor with at least two observers; tails=selective shunt). All operations were performed by one of three surgeons (B. L., B. R. T., and R. M. S.) using the standard technique described by Warren for the distal shunt, except that in a few patients the renal vein was divided for technical reasons and an end-to-end splenorenal anastomosis was constructed. The control operation in all cases was the end-to-side portacaval shunt. Anesthetic, operative techniques, and postoperative management were similar in all cases. Early selective shunt patency was studied in 27 patients: 22 by angiography, 4 at autopsy, and 1 at a subsequent operation. Portacaval shunt patency was not routinely studied. Regular follow-up was carried out in the Toronto General Hospital Shunt Clinic by the medical-surgical team or, where distance precluded that, by communication with the referring physician and semiannual reviews. All data
DISTAL
SPLENORENAL
SHUNT TRIAL
425
were obtained in standardized fashion by nonblinded observers and recorded on computer. Evidence of encephalopathy was sought at regular clinic follow-up or on readmission to hospital. The diagnosis was based on careful history, physical examination, trail testing, and, in symptomatic patients, electroencephalogram. Significant symptoms included neuropsychiatric abnormalities reported by the patient or his family which were not explained by stroke, intoxication, or other disorder. Signs included asterixis, gait disorders, and impaired mentation on testing unless these were clearly due to other neurologic disease. We termed encephalopathy “induced” if it was precipitated by gastrointestinal bleeding or acute binge drinking, or was part of the picture of terminal hepatic failure. Other episodes of encephalopathy were considered “spontaneous.” We termed encephalopathy “severe” if it precluded normal activities of daily living, caused at least one hospital admission, or required prolonged medical therapy. “Mild” encephalopathy was that easily managed by diet, not interfering with normal function, and never requiring hospitalization. Patients without impairment, who were fully employed or engaged in their normal household tasks, were scored “no disability.” Patients with moderate interference with their life-style including modification of their work performance or partial limitation of household and self-care ability were called “mild disability.” Those in the “severe disability” classification were patients who were unable to work and who required assistance for the basic activities of daily living. Patients were reviewed monthly in the early postoperative period and follow-up was then extended to at least twice yearly in long survivors. At follow-up, an assessment of functional status in each of the preceding months was made and recorded as described previously. Any period of “mild disability” or “severe disability” in any month was counted as 1 mo in that functional category. Continuous data are reported as mean t- SEM. Statistical comparisons were made by x’, Student’s t-test, Fischer’s exact test, and standard life-table analysis as appropriate.
Results Eighty patients were entered into the trial between 1971 and 1981, representing -30% of the shunt procedures performed at our institution over this time period. Complex and changing referral patterns for variceal bleeding in our tertiary care center unfortunately make any calculation of accession rates arbitrary. Forty-one patients were randomized to the portacaval shunt group and 39 to the selective shunt group. Technical considerations during the procedure led to exclusion of 1 patient randomized to selective shunt and 1 patient to portacaval shunt. These 2 patients have been excluded from the following data analysis although they have both done well postoperatively. There was no significant difference between the two shunt groups with regard to age, sex distribu-
426 LANGER ET AL.
Table
1. Study
GASTROENTEROLOGY Vol.88.No. 2
operation (three selective and five portacaval operations). Selective shunt operations required more time than portacaval operations (mean 3.9 -+ 0.2 h vs. 2.8 + 0.1 h, p < 0.01) and also more blood replacement (mean 4.6 k 0.6 U vs. 2.5 t 0.2 U, p < O.Ol), confirming the increased technical difficulty of that procedure. There were, however, no intraoperative deaths in either group. Early selective shunt patency, based on angiography or autopsy, was 88%. Loss of portal perfusion was demonstrated only in the 2 patients who developed portal vein thrombosis. The operative mortality rate, that is, death while still in hospital or within 30 days of operation, was 0 for the portacaval group and 5 of 38 (13%) for the selective group (Table 2), although only 2 of these patients died within 30 days of operation. One patient died of thrombosis of the portal vein and a clotted shunt, 1 died of a clotted shunt only, and 3 died of hepatic failure with patent shunts and portal veins. Our initial report indicated a significant difference in mortality between portacaval shunt (0 of 28 deaths) and distal shunt (5 of 27, p < 0.05). Continued experience with the distal shunt has reduced the postoperative complication and mortality rate. There have been no postoperative deaths in any of the 31 patients entered into the trial since early 1977. The presently reported overall difference in operative mortality remains significant (p = 0.024 FE5), but there is a trend in postoperative mortality rate of distal splenorenal shunt toward equivalence with portacaval shunt as a result of increasing experience with the operation. There was no difference in the early (in hospital) postoperative complication rates between the two groups (Table 3). The only confirmed variceal bleed was in 1 selective shunt patient. Postoperative encephalopathy occurred in both groups of patients and was reversible except in those patients who died of hepatic failure. Postoperative ascites was common in both groups and was usually managed satisfactorily by medical means. Eleven patients had persistent tense ascites. In 3 of the 11, frank hepatorenal syndrome devel-
Patients
Age, mean (range) Sex, % male Pathology Alcoholic cirrhosis Nonalcoholic cirrhosis Chronic active hepatitis Primary biliary cirrhosis Cryptogenic History of prior bleed History of prior encephalopathy Child class A B C Child parameters Bilirubin, total (mgidl 2 SE)
Selective
Portacaval
(n=38)
(n=40)
49 (19-68)
50 (25-67) 70
66 32
32
3 1 2 33 9
3 1 4 32 7
14 20 (3)” 4
11 25 (5)” 4
2.43 k 0.4 3.13 5 0.07
Albumin (gidl) Ascites Mild Severe Encephalopathy Mild Severe Nutrition Good or fair Poor
2.56 +- 0.5 3.22 t 0.07
9 4
8 3
2 2
0 0
37 1
38 2
’ Parentheses indicate number of patients who were Child class C on admission, but improved to Child class B by the time of their operation.
tion, pathology, history of previous bleeding or encephalopathy, preoperative Child classification, or any of Child’s individual parameters (Table 1). Most of our patients were alcoholic cirrhotics, as judged by history and liver histology. Unlike our overall variceal bleeder population, the trial includes mainly patients whose clinical staging was Child classification A or B. Many of the Child C patients in our practice required emergency surgical procedures or were excluded from this trial because of severe alcoholic hepatitis on liver biopsy, severe encephalopathy, or uncontrollable ascites. However, 8 patients classified as Child class C at the time of their bleeding improved with supportive care to Child class B at the time of randomization and Table
2. Early (in Hospital)
Deathsa Operation
Year
Age (yr)
1971 1972 1973 1974
68 39 30 53
1977
63
a Five selective;
no portacaval.
Alcohol + + +
Child class
Blood
Time
WI
(hl
B C A A
7 19 5 9
4 7 5 5
B
1
3
Survival Cause
of death
Hepatic failure Hepatorenal syndrome Hepatic failure Shunt & portal vein clot Shunt clot
(days) 2.5 21 55 41 35
February
1985
DISTAL
Table 3. Early [in Hospital) Complications Portacaval (n=40)
Selective (n=38) Encephalopathy
7
4
Ascites Mild Tense Acute renal failure Hepatorenal syndrome Portal vein thrombosis
11 5 4 2 2
6 6 0 1 0
Gastrointestinal bleed Varices Other site Unknown site Pneumonia Wound infection Other
1 1 2 3 2 7
0 1 0 2 2 8
oped and was fatal in 2 of the 3. Acute renal tubular necrosis developed in 4 shunt patients and was associated with fatal outcomes in 2 patients. Since 1977, we have been using the peritoneovenous shunt to manage acute postoperative intractable ascites. We have in this way successfully managed 1 selective and 2 portacaval shunt patients in the trial, as well as a number of others not entered in the study. Long-term follow-up was complete in 95% of selective shunt patients and in 100% of portacaval shunt patients. Mean follow-up time from operation to last data analysis (both alive and dead patients) was 65 mo in both groups (range 4-129 mo). Late complications (Table 4) included upper gastrointestinal bleeding, which occurred in 4 portacaval shunt patients and 5 selective shunt patients. Only 1 of the 5 was a confirmed variceal bleed (selective). Bleeding in other patients was from ulcer (z), gastritis (2), or undiagnosed (5); overall rebleeding rates were not significantly different. The incidence of induced encephalopathy after portacaval shunt (8%) was not significantly different
SPLENORENAL
SHlJNT
Selective (n=33)
Gastrointestinal Variceal Other site Unknown site Total Encephalopathy Induced Spontaneous Mild Severe Total cl p < 0.05.
Portacaval (n=40)
No.
“/1
No.
%
1 2 2 5
3 6 6 15
0 2 2
0 5 5
4
10
4
12
3
8
bleeding
4
12
3
4
12
13
33”
8
8
24
16
41
427
from that after selective shunt (12%, Table 4). Spontaneous encephalopathy, however, occurred significantly more often after portacaval shunt than after selective shunt (40% vs. 24%, p L 0.12). Significantly, severe encephalopathy troubled 33% of portacaval shunt patients but only 12% of selective shunt patients (p = 0.036). The burden of this metabolic problem is reflected in admission to hospital. Encephalopathy precipitated 26 late admissions in 13 portacaval shunt patients but only six late admissions in 4 selective shunt patients (Table 5). Thus the risk of admission for encephalopathy was more than three times higher in the portacaval group (0.65 vs. 0.18 admissions per patient, p < 0.001). As these admissions accumulate after surgery (Figure l), portacaval patients appear at a disadvantage throughout their course, and not just early or late. Overall late hospital admissions, another index of quality of life, were more than twice as frequent in the portacaval group (57 vs. 22 admissions, or 1.43 vs. 0.65 admissions per patient, p < 0.01, Table 5). Functional status scores are reported in Table 5 as patient-months spent in each of the three disability classes. Selective shunt patients remained without functional disability significantly longer on average than portacaval shunt patients (83% vs. 70% of patient-months, p < 0.05). Furthermore, disability in the selective shunt group, when it occurred, was less severe (1% vs. 5% of patient-years in severe disability class, p < 0.05). Long-term survival as projected by actuarial analysis (Figure 2) was not significantly different in the two groups whether or not the postoperative deaths were included. If they were included, 5-yr survival rates were 44% (selective) vs. 56% (portacaval); if they were not included, they were 51% and 56%, Table 5. Assessment of Oualitv of Life Selfxtive (no= 33)
Table 4. Late Complications
TRIAL
Admissions for encephalopathy Admissions Patients admitted Risk (admissions/patient)
6 4 (12”/i;,) 0.18 2 0.5 SD
Admissions for all causes Admissions Patients admitted Risk (admissions/patient]
Functional status (patient-months] No disability Mild disabilitv Severe disability u p < 0.001.
” p < 0.05.
Portacaval (n=40)
26 13 (33%) 0.65
+ 1.18 SD’
57
22 13 [39%,)
22 (55%)
0.67
1.43
+ 0.9 SD
? 1.7 SD
No.
%
No.
1127
83
1128
70
223 13
16 1
396 84
25 5”
‘X1
428
LANGER
ET AL.
GASTROENTEROLOGY
SURVIVAL,
%
I---.
ENCEPHALOf’ATHY
.~
Vol.
88, No.
2
SELECTIVE -. PORTACAVAL
PORTACAVAL ‘~
I
10
Figure
6
3
2. Actuarial analysis of survival, ative deaths (n = 38 selective,
9 YEARS
including early postopern = 40 portacaval].
YEARS 1
Figure
2
3
4
5
1. Cumulative hospital admissions for encephalopathy in the 33 selective shunt and 40 portacaval shunt postoperative survivors, up to 5 yr after surgery.
respectively. The trend that appeared to favor the selective shunt in our preliminary report is no longer apparent. Hepatic failure was the most common cause of late death in both groups (Table 6). Persistent alcohol abuse was a significant factor in three of five selective shunt hepatic failure deaths but in only one of eight such portacaval deaths. Alcohol was a contributing factor in an additional two selective shunt and two portacaval shunt late deaths. Recurrent gastrointestinal bleeding was not a prominent cause of late death in either group, confirming the value of both shunts in preventing rebleeding. We found no preoperative risk factor which, singly or in combination, reliably correlated with either early or late mortality (i.e., age, prior bleeds or encephalopathy, Child class, hemodynamics, pathology). In particular, the few nonalcoholic cirrhotics in the study (14 postoperative survivors) lived no longer than did alcoholic cirrhotics. Encephalopathy and survival rates in our small nonalcoholic selective shunt group (n = 5) were not significantly different from our nonalcoholic portacaval shunt patients (n = 9).
Discussion The end-to-side portacaval shunt has been the standard operation for portosystemic decompression since the 1940s and there is more information available about this operation than any other. Whether or not the mesocaval shunt is comparable in effectiveness, ease, patency, or safety remains an open question, and for this reason our trial restricted the control operation to standard end-to-side portacaval shunt with ligation of proximal portal vein. The two patient groups in our randomized trial displayed similar clinical and demographic findings preoperatively. All operations were performed by one of three surgeons participating in the trial who by using
similar techniques assured uniformity. Ancillary preoperative and postoperative care as well as anesthetic technique was similar in both groups of patients. The mortality rate in our first report was significantly higher in the selective shunt group (5 of 28 patients). After an additional 24 patient entries, this difference has not dropped below statistical signihcance. However, the operative mortality of portacaval shunt in our trial is unexpectedly low (0%) and tends to magnify the difference between it and the operative mortality of selective shunt. The other reported trials (7-9) have reported no significant difference in mortality between operations, although the portacaval shunt mortality rate in one trial (9) is higher than that generally reported for elective shunts. Three of our postoperative deaths followed prolonged operations associated with excessive blood loss in our early experience. In retrospect, these procedures should have been abandoned in favor of easier total shunts. The distal splenorenal shunt can be a difficult, time-consuming, technical procedure which probably should not be undertaken by surgeons who perform less than one shunt per month. Postshunt encephalopathy in our study was significantly less frequent after selective than after portacaval shunt. This is in agreement with the studies of Galambos et al. (7) and Reichle et al. (8); however, the report of Conn et al. (9) failed to show any Table
6.
Causes
Hepatic failure Respiratory Trauma Gastrointestinal Cardiac Hepatoma Astrocytoma Unknown a Parentheses
of Late
hemorrhage
indicate
alcohol
Death Selective (n=151
Portacaval (n=191
5 (3)” 2 2 (21 2 0 2 0 2
8 (11 3 (1) 1 (1) 1 2 0 1 3
abuse-related
death.
February1985
difference in encephalopathy rates between selective and total shunts. It should be noted, however, that in this last study there was less standardization of either the test or control operation, in that there were 10 different surgeons, each doing a mean of 2.4 distal shunts, and 2.9 total shunts (mesocaval, end-to-side, portacaval, or side-to-side portacaval). In addition, that report did not include follow-up angiographic data to demonstrate preservation of hepatopetal flow after distal shunt, which is felt to be a critical factor in the protection against postshunt encephalopathy. In our study the encephalopathy rate seems to correlate well with the long-term quality of life, which is clearly superior in distal shunt patients to that following end-to-side portacaval shunt. Selective shunt survivors were rehospitalized at less than half the rate of their portacaval shunt counterparts. Encephalopathy was not only less frequent, but also less severe. Functional status throughout the followup period was demonstrably superior. In spite of these differences, however, we cannot demonstrate any difference in survival between the two groups, or any correlation between postoperative encephalopathy and late mortality. Uncontrolled studies suggest that nonalcoholics enjoy improved survival after selective shunt compared with portacaval shunt (10). We cannot confirm this hypothesis in our small nonalcoholic group. Continued alcohol abuse, however, contributed to more late deaths from liver failure in our selective group than in our portacaval shunt group. The gradual development of collateral circulation in selective shunt patients has been described by Widrich et al. (12), Maillard et al. (13), and others. Theoretically, this loss of portal flow might negate the beneficial effects of the selective shunt. If this were true, one would expect to see encephalopathy occurring in these patients and eventually becoming equal to the total shunt group. This has not occurred thus far in our study. It may be that patients who gradually develop collaterals and slowly lose their portal inflow fare better than those whose portal flow is suddenly diverted at the time of total shunting. Further follow-up with angiography or flow studies, or both, may clarify this point. We have concluded from this trial that the distal splenorenal shunt is a technically more difficult
DISTAL
SPLENORENAL
SHUNT
TRIAL
429
operation than the portacaval shunt and is associated with a higher postoperative mortality rate in the learning phase. In experienced hands, it can be carried out with an operative mortality similar to portacaval shunt. It is as effective as portacaval shunt in preventing rebleeding. It is associated with a lower incidence of postoperative encephalopathy, particularly severe postshunt encephalopathy, and a better quality of life. However, improvement in longterm survival following this shunt as compared with portacaval shunt has not been demonstrated.
References 1. Warren
WD, Zeppa R, Fomon JJ. Selective trans-splenic decompression of gastroesophageal varices by distal splenorenal shunt. Ann Surg 1967;166:437-55, 2. Warren WD, Rudman D, Millikan W, et al. The metabolic basis of portasystemic encephalopathy and the effect of selective versus nonselective shunts. Ann Surg 1974;180:573-9. 3. Hutson DG, Pereiras R, Zeppa R, et al. The fate of esophageal varices following selective distal splenorenal shunt. Ann Surg 1976;183:496-501. 4. Mosimann R, Loup P. Efficacy and risks of the distal splenorenal shunt in the treatment of bleeding esophageal varices. Am J Surg 1977;133:163-8. 5. Silver DS, Puckett CL, McNeer JF, et al. Evaluation of selective transsplenic decompression of gastroesophageal varices. Am J Surg 1974;127:30-4. 6. Zeppa R, Hutson DG, Bergstresser PR, et al. Survival after distal splenorenal shunt. Surg Gynecol Obstet 1977;145:12-6. 7. Galambos JT, Warren WD, Rudman D. et al. Selective and total shunts in the treatment of bleeding varices. A randomized controlled trial. N Engl J Med 1976;295:1089-95. a. Reichle FA, Fahmy WF, Golsorkhi M. Prospective comparative clinical trial with distal splenorenal and mesocaval shunts. Am J Surg 1979;137:13-21. 9. Conn HO, Resnick RH, Grace ND, et al. Distal splenorenal shunt versus portal-systemic shunt: current status of a controlled trial. Hepatology 1981;1:151-60. 10. Zeppa R, Hensley GT, Levi JU, et al. The comparative survivals of alcoholics versus nonalcoholics after distal splenorenal shunt. Ann Surg 1978;187:510-4. 11. Langer B, Stone RM, Pate1 SC, Colapinto RF. A prospective randomized trial of the selective distal splenorenal shunt. Surg Gynecol Obstet 1980;150:45-8. 12. Widrich WC, Robbins AH, Nabseth DC, et al. Portal hypertension changes following selective splenorenal shunt surgery: evaluation by percutaneous transhepatic portal catheterization, venography, and cinefluorography. Radio 1976:121:
295-302. 13. Maillard JN, Flamant the distal
splenorenal
YM, Hay JM, Chandler JG. Selectivity shunt. Surgery 1979:86:663-71.
of
1985;88:424-9
Further Report of a Prospective Randomized Trial Comparing Distal Splenorenal Shunt With End-to-Side Portacaval Shunt An Analysis of Encephalopathy, Quality of Life
Survival, and
B. LANGER, B. R. TAYLOR, D. R. MACKENZIE, R. M. STONE, and L. BLENDIS Department of Surgery, University of Toronto: General Hospital; and Departments of Surgery, Hospital, Toronto, Ontario, Canada
We electively compared the distal splenorenal (“selective”) shunt with the end-to-side portacaval shunt in 80 prospectively randomized patients with variceal bleeding. Selective shunts required more operative time (3.9 vs. 2.8 h) and blood replacement (4.6 vs. 2.5 LJ) and postoperative mortality was slightly higher (5 of 38 selective vs. 0 of 40 portacaval). Postoperative complication rates were similar. After 65-mo mean follow-up, both shunts have protected well against late gastrointestinal bleeding (5 selective, 4 portacaval episodes). However, after selective shunts, spontaneous encephalopathy occurred less often (23% vs. 40% of patients), was severe in fewer patients (12% vs. 33%), and precipitated fewer hospital admissions (6 admissions in 4 selective patients vs. 26 admissions in 13 portacaval patients). Furthermore, selective shunt patients remained longer without functional disability (83% vs. 70% of postoperative patient months). Long-term survival was not significantly different in the two groups (5yr survival: selective 51% portacaval 56%). In 1967, Warren and colleagues concept of selective decompression
(1)
introduced the of esophagogas-
Received February 24, 1983. Accepted August 16, 1984. Address requests for reprints to: B. Langer, M.D., Toronto General Hospital, Eaton Building N, 9-236, Toronto, Canada M5G lL7. This work was supported by the General Surgery Research Fund, Toronto General Hospital. The authors thank Denise Reynolds for manuscript preparation. 0 1985 by the American Gastroenterological Association 0016.5085/85/$3.30
T. GILAS,
Departments of Surgery and Medicine, Toronto Western Hospital and Mount
Toronto Sinai
tric varices for patients with portal hypertension and bleeding. This procedure was developed to avoid complete portal blood diversion, which occurs after standard total shunt procedures such as portacaval or mesocaval shunts. This total diversion was believed to be the major contributing factor in the development of postshunt encephalopathy (PSE) and late mortality (2). The effectiveness of the selective shunt in controlling bleeding has been documented in several reports (3-5). The technical difficulty in its performance has also been noted. With experience, however, the operative morbidity and mortality can be as low as that of standard total shunts (6). A lower PSE incidence following the selective shunt has been documented in two controlled trials from other centers (7,8]. The third trial (9) failed to show any difference in encephalopathy rates between selective and total shunts. An improvement in long-term survival has been claimed in nonalcoholics based on uncontrolled data (10); however, improved survival has not yet been shown in any of the controlled trials (7-9). In 1971, we began a prospective randomized trial comparing the distal splenorenal shunt (“selective”] with the end-to-side portacaval shunt (“portacaval”). A preliminary report of this elective, therapeutic trial in 1980 (11) stated that the operative mortality in the selective shunt was higher than in the portacaval shunt, but the postshunt encephalopathy Abbreviation athy.
used
in this
paper:
PSE, postshunt
encephalop-
February
1985
rate was significantly lower. There was then no difference in long-term survival rates, although the data suggested a trend favoring the selective shunt. This report is a further follow-up of the 56 patients originally reported and an additional 24 patients who have been entered until January 1981 and evaluated until April 1982.
Methods As reported previously, all patients in the trial were biopsy-proven cirrhotics with portal hypertension who had survived at least one major variceal hemorrhage (requiring a minimum 3-U transfusion at our institutions or elsewhere). Actively bleeding patients received intensive care support under a combined medical-surgical team. Nonoperative treatment, including vasopressin and balloon tamponade, was used to control bleeding in the first 24 h. If bleeding continued or recurred, and the patient was considered operable, an emergency portacaval or mesocaval shunt was performed. If the bleeding was controlled medically, patients were stabilized and then electively considered for entry into the distal shunt trial. Jaundice, previous encephalopathy, and controlled ascites were not contraindications to entry into the trial. Endoscopy, liver biopsies, and angiography were carried out in all patients before entry to document the nature of liver disease, to confirm that there were no other upper gastrointestinal tract lesions to account for the bleeding, and to demonstrate that splenic and renal veins were suitable for shunting. Patients were excluded if they refused randomization, were over 70 yr of age, had hepatofugal portal flow, or had severe alcoholic hepatitis by histologic criteria (>lO% of cells with hyaline) or on clinical grounds (fever, leukocystosis, and tender liver). Also excluded were those with associated lesions requiring simultaneous treatment (peptic ulcer or hiatus hernia], severe uncontrollable ascites, severe coagulopathy [refractory to administration of clotting factors), or persistent spontaneous encephalopathy. If such patients later improved, they were again considered for the trial. Written informed consent was obtained and randomization was carried out in the operating room by surgeon’s coin toss (to the floor with at least two observers; tails=selective shunt). All operations were performed by one of three surgeons (B. L., B. R. T., and R. M. S.) using the standard technique described by Warren for the distal shunt, except that in a few patients the renal vein was divided for technical reasons and an end-to-end splenorenal anastomosis was constructed. The control operation in all cases was the end-to-side portacaval shunt. Anesthetic, operative techniques, and postoperative management were similar in all cases. Early selective shunt patency was studied in 27 patients: 22 by angiography, 4 at autopsy, and 1 at a subsequent operation. Portacaval shunt patency was not routinely studied. Regular follow-up was carried out in the Toronto General Hospital Shunt Clinic by the medical-surgical team or, where distance precluded that, by communication with the referring physician and semiannual reviews. All data
DISTAL
SPLENORENAL
SHUNT TRIAL
425
were obtained in standardized fashion by nonblinded observers and recorded on computer. Evidence of encephalopathy was sought at regular clinic follow-up or on readmission to hospital. The diagnosis was based on careful history, physical examination, trail testing, and, in symptomatic patients, electroencephalogram. Significant symptoms included neuropsychiatric abnormalities reported by the patient or his family which were not explained by stroke, intoxication, or other disorder. Signs included asterixis, gait disorders, and impaired mentation on testing unless these were clearly due to other neurologic disease. We termed encephalopathy “induced” if it was precipitated by gastrointestinal bleeding or acute binge drinking, or was part of the picture of terminal hepatic failure. Other episodes of encephalopathy were considered “spontaneous.” We termed encephalopathy “severe” if it precluded normal activities of daily living, caused at least one hospital admission, or required prolonged medical therapy. “Mild” encephalopathy was that easily managed by diet, not interfering with normal function, and never requiring hospitalization. Patients without impairment, who were fully employed or engaged in their normal household tasks, were scored “no disability.” Patients with moderate interference with their life-style including modification of their work performance or partial limitation of household and self-care ability were called “mild disability.” Those in the “severe disability” classification were patients who were unable to work and who required assistance for the basic activities of daily living. Patients were reviewed monthly in the early postoperative period and follow-up was then extended to at least twice yearly in long survivors. At follow-up, an assessment of functional status in each of the preceding months was made and recorded as described previously. Any period of “mild disability” or “severe disability” in any month was counted as 1 mo in that functional category. Continuous data are reported as mean t- SEM. Statistical comparisons were made by x’, Student’s t-test, Fischer’s exact test, and standard life-table analysis as appropriate.
Results Eighty patients were entered into the trial between 1971 and 1981, representing -30% of the shunt procedures performed at our institution over this time period. Complex and changing referral patterns for variceal bleeding in our tertiary care center unfortunately make any calculation of accession rates arbitrary. Forty-one patients were randomized to the portacaval shunt group and 39 to the selective shunt group. Technical considerations during the procedure led to exclusion of 1 patient randomized to selective shunt and 1 patient to portacaval shunt. These 2 patients have been excluded from the following data analysis although they have both done well postoperatively. There was no significant difference between the two shunt groups with regard to age, sex distribu-
426 LANGER ET AL.
Table
1. Study
GASTROENTEROLOGY Vol.88.No. 2
operation (three selective and five portacaval operations). Selective shunt operations required more time than portacaval operations (mean 3.9 -+ 0.2 h vs. 2.8 + 0.1 h, p < 0.01) and also more blood replacement (mean 4.6 k 0.6 U vs. 2.5 t 0.2 U, p < O.Ol), confirming the increased technical difficulty of that procedure. There were, however, no intraoperative deaths in either group. Early selective shunt patency, based on angiography or autopsy, was 88%. Loss of portal perfusion was demonstrated only in the 2 patients who developed portal vein thrombosis. The operative mortality rate, that is, death while still in hospital or within 30 days of operation, was 0 for the portacaval group and 5 of 38 (13%) for the selective group (Table 2), although only 2 of these patients died within 30 days of operation. One patient died of thrombosis of the portal vein and a clotted shunt, 1 died of a clotted shunt only, and 3 died of hepatic failure with patent shunts and portal veins. Our initial report indicated a significant difference in mortality between portacaval shunt (0 of 28 deaths) and distal shunt (5 of 27, p < 0.05). Continued experience with the distal shunt has reduced the postoperative complication and mortality rate. There have been no postoperative deaths in any of the 31 patients entered into the trial since early 1977. The presently reported overall difference in operative mortality remains significant (p = 0.024 FE5), but there is a trend in postoperative mortality rate of distal splenorenal shunt toward equivalence with portacaval shunt as a result of increasing experience with the operation. There was no difference in the early (in hospital) postoperative complication rates between the two groups (Table 3). The only confirmed variceal bleed was in 1 selective shunt patient. Postoperative encephalopathy occurred in both groups of patients and was reversible except in those patients who died of hepatic failure. Postoperative ascites was common in both groups and was usually managed satisfactorily by medical means. Eleven patients had persistent tense ascites. In 3 of the 11, frank hepatorenal syndrome devel-
Patients
Age, mean (range) Sex, % male Pathology Alcoholic cirrhosis Nonalcoholic cirrhosis Chronic active hepatitis Primary biliary cirrhosis Cryptogenic History of prior bleed History of prior encephalopathy Child class A B C Child parameters Bilirubin, total (mgidl 2 SE)
Selective
Portacaval
(n=38)
(n=40)
49 (19-68)
50 (25-67) 70
66 32
32
3 1 2 33 9
3 1 4 32 7
14 20 (3)” 4
11 25 (5)” 4
2.43 k 0.4 3.13 5 0.07
Albumin (gidl) Ascites Mild Severe Encephalopathy Mild Severe Nutrition Good or fair Poor
2.56 +- 0.5 3.22 t 0.07
9 4
8 3
2 2
0 0
37 1
38 2
’ Parentheses indicate number of patients who were Child class C on admission, but improved to Child class B by the time of their operation.
tion, pathology, history of previous bleeding or encephalopathy, preoperative Child classification, or any of Child’s individual parameters (Table 1). Most of our patients were alcoholic cirrhotics, as judged by history and liver histology. Unlike our overall variceal bleeder population, the trial includes mainly patients whose clinical staging was Child classification A or B. Many of the Child C patients in our practice required emergency surgical procedures or were excluded from this trial because of severe alcoholic hepatitis on liver biopsy, severe encephalopathy, or uncontrollable ascites. However, 8 patients classified as Child class C at the time of their bleeding improved with supportive care to Child class B at the time of randomization and Table
2. Early (in Hospital)
Deathsa Operation
Year
Age (yr)
1971 1972 1973 1974
68 39 30 53
1977
63
a Five selective;
no portacaval.
Alcohol + + +
Child class
Blood
Time
WI
(hl
B C A A
7 19 5 9
4 7 5 5
B
1
3
Survival Cause
of death
Hepatic failure Hepatorenal syndrome Hepatic failure Shunt & portal vein clot Shunt clot
(days) 2.5 21 55 41 35
February
1985
DISTAL
Table 3. Early [in Hospital) Complications Portacaval (n=40)
Selective (n=38) Encephalopathy
7
4
Ascites Mild Tense Acute renal failure Hepatorenal syndrome Portal vein thrombosis
11 5 4 2 2
6 6 0 1 0
Gastrointestinal bleed Varices Other site Unknown site Pneumonia Wound infection Other
1 1 2 3 2 7
0 1 0 2 2 8
oped and was fatal in 2 of the 3. Acute renal tubular necrosis developed in 4 shunt patients and was associated with fatal outcomes in 2 patients. Since 1977, we have been using the peritoneovenous shunt to manage acute postoperative intractable ascites. We have in this way successfully managed 1 selective and 2 portacaval shunt patients in the trial, as well as a number of others not entered in the study. Long-term follow-up was complete in 95% of selective shunt patients and in 100% of portacaval shunt patients. Mean follow-up time from operation to last data analysis (both alive and dead patients) was 65 mo in both groups (range 4-129 mo). Late complications (Table 4) included upper gastrointestinal bleeding, which occurred in 4 portacaval shunt patients and 5 selective shunt patients. Only 1 of the 5 was a confirmed variceal bleed (selective). Bleeding in other patients was from ulcer (z), gastritis (2), or undiagnosed (5); overall rebleeding rates were not significantly different. The incidence of induced encephalopathy after portacaval shunt (8%) was not significantly different
SPLENORENAL
SHlJNT
Selective (n=33)
Gastrointestinal Variceal Other site Unknown site Total Encephalopathy Induced Spontaneous Mild Severe Total cl p < 0.05.
Portacaval (n=40)
No.
“/1
No.
%
1 2 2 5
3 6 6 15
0 2 2
0 5 5
4
10
4
12
3
8
bleeding
4
12
3
4
12
13
33”
8
8
24
16
41
427
from that after selective shunt (12%, Table 4). Spontaneous encephalopathy, however, occurred significantly more often after portacaval shunt than after selective shunt (40% vs. 24%, p L 0.12). Significantly, severe encephalopathy troubled 33% of portacaval shunt patients but only 12% of selective shunt patients (p = 0.036). The burden of this metabolic problem is reflected in admission to hospital. Encephalopathy precipitated 26 late admissions in 13 portacaval shunt patients but only six late admissions in 4 selective shunt patients (Table 5). Thus the risk of admission for encephalopathy was more than three times higher in the portacaval group (0.65 vs. 0.18 admissions per patient, p < 0.001). As these admissions accumulate after surgery (Figure l), portacaval patients appear at a disadvantage throughout their course, and not just early or late. Overall late hospital admissions, another index of quality of life, were more than twice as frequent in the portacaval group (57 vs. 22 admissions, or 1.43 vs. 0.65 admissions per patient, p < 0.01, Table 5). Functional status scores are reported in Table 5 as patient-months spent in each of the three disability classes. Selective shunt patients remained without functional disability significantly longer on average than portacaval shunt patients (83% vs. 70% of patient-months, p < 0.05). Furthermore, disability in the selective shunt group, when it occurred, was less severe (1% vs. 5% of patient-years in severe disability class, p < 0.05). Long-term survival as projected by actuarial analysis (Figure 2) was not significantly different in the two groups whether or not the postoperative deaths were included. If they were included, 5-yr survival rates were 44% (selective) vs. 56% (portacaval); if they were not included, they were 51% and 56%, Table 5. Assessment of Oualitv of Life Selfxtive (no= 33)
Table 4. Late Complications
TRIAL
Admissions for encephalopathy Admissions Patients admitted Risk (admissions/patient)
6 4 (12”/i;,) 0.18 2 0.5 SD
Admissions for all causes Admissions Patients admitted Risk (admissions/patient]
Functional status (patient-months] No disability Mild disabilitv Severe disability u p < 0.001.
” p < 0.05.
Portacaval (n=40)
26 13 (33%) 0.65
+ 1.18 SD’
57
22 13 [39%,)
22 (55%)
0.67
1.43
+ 0.9 SD
? 1.7 SD
No.
%
No.
1127
83
1128
70
223 13
16 1
396 84
25 5”
‘X1
428
LANGER
ET AL.
GASTROENTEROLOGY
SURVIVAL,
%
I---.
ENCEPHALOf’ATHY
.~
Vol.
88, No.
2
SELECTIVE -. PORTACAVAL
PORTACAVAL ‘~
I
10
Figure
6
3
2. Actuarial analysis of survival, ative deaths (n = 38 selective,
9 YEARS
including early postopern = 40 portacaval].
YEARS 1
Figure
2
3
4
5
1. Cumulative hospital admissions for encephalopathy in the 33 selective shunt and 40 portacaval shunt postoperative survivors, up to 5 yr after surgery.
respectively. The trend that appeared to favor the selective shunt in our preliminary report is no longer apparent. Hepatic failure was the most common cause of late death in both groups (Table 6). Persistent alcohol abuse was a significant factor in three of five selective shunt hepatic failure deaths but in only one of eight such portacaval deaths. Alcohol was a contributing factor in an additional two selective shunt and two portacaval shunt late deaths. Recurrent gastrointestinal bleeding was not a prominent cause of late death in either group, confirming the value of both shunts in preventing rebleeding. We found no preoperative risk factor which, singly or in combination, reliably correlated with either early or late mortality (i.e., age, prior bleeds or encephalopathy, Child class, hemodynamics, pathology). In particular, the few nonalcoholic cirrhotics in the study (14 postoperative survivors) lived no longer than did alcoholic cirrhotics. Encephalopathy and survival rates in our small nonalcoholic selective shunt group (n = 5) were not significantly different from our nonalcoholic portacaval shunt patients (n = 9).
Discussion The end-to-side portacaval shunt has been the standard operation for portosystemic decompression since the 1940s and there is more information available about this operation than any other. Whether or not the mesocaval shunt is comparable in effectiveness, ease, patency, or safety remains an open question, and for this reason our trial restricted the control operation to standard end-to-side portacaval shunt with ligation of proximal portal vein. The two patient groups in our randomized trial displayed similar clinical and demographic findings preoperatively. All operations were performed by one of three surgeons participating in the trial who by using
similar techniques assured uniformity. Ancillary preoperative and postoperative care as well as anesthetic technique was similar in both groups of patients. The mortality rate in our first report was significantly higher in the selective shunt group (5 of 28 patients). After an additional 24 patient entries, this difference has not dropped below statistical signihcance. However, the operative mortality of portacaval shunt in our trial is unexpectedly low (0%) and tends to magnify the difference between it and the operative mortality of selective shunt. The other reported trials (7-9) have reported no significant difference in mortality between operations, although the portacaval shunt mortality rate in one trial (9) is higher than that generally reported for elective shunts. Three of our postoperative deaths followed prolonged operations associated with excessive blood loss in our early experience. In retrospect, these procedures should have been abandoned in favor of easier total shunts. The distal splenorenal shunt can be a difficult, time-consuming, technical procedure which probably should not be undertaken by surgeons who perform less than one shunt per month. Postshunt encephalopathy in our study was significantly less frequent after selective than after portacaval shunt. This is in agreement with the studies of Galambos et al. (7) and Reichle et al. (8); however, the report of Conn et al. (9) failed to show any Table
6.
Causes
Hepatic failure Respiratory Trauma Gastrointestinal Cardiac Hepatoma Astrocytoma Unknown a Parentheses
of Late
hemorrhage
indicate
alcohol
Death Selective (n=151
Portacaval (n=191
5 (3)” 2 2 (21 2 0 2 0 2
8 (11 3 (1) 1 (1) 1 2 0 1 3
abuse-related
death.
February1985
difference in encephalopathy rates between selective and total shunts. It should be noted, however, that in this last study there was less standardization of either the test or control operation, in that there were 10 different surgeons, each doing a mean of 2.4 distal shunts, and 2.9 total shunts (mesocaval, end-to-side, portacaval, or side-to-side portacaval). In addition, that report did not include follow-up angiographic data to demonstrate preservation of hepatopetal flow after distal shunt, which is felt to be a critical factor in the protection against postshunt encephalopathy. In our study the encephalopathy rate seems to correlate well with the long-term quality of life, which is clearly superior in distal shunt patients to that following end-to-side portacaval shunt. Selective shunt survivors were rehospitalized at less than half the rate of their portacaval shunt counterparts. Encephalopathy was not only less frequent, but also less severe. Functional status throughout the followup period was demonstrably superior. In spite of these differences, however, we cannot demonstrate any difference in survival between the two groups, or any correlation between postoperative encephalopathy and late mortality. Uncontrolled studies suggest that nonalcoholics enjoy improved survival after selective shunt compared with portacaval shunt (10). We cannot confirm this hypothesis in our small nonalcoholic group. Continued alcohol abuse, however, contributed to more late deaths from liver failure in our selective group than in our portacaval shunt group. The gradual development of collateral circulation in selective shunt patients has been described by Widrich et al. (12), Maillard et al. (13), and others. Theoretically, this loss of portal flow might negate the beneficial effects of the selective shunt. If this were true, one would expect to see encephalopathy occurring in these patients and eventually becoming equal to the total shunt group. This has not occurred thus far in our study. It may be that patients who gradually develop collaterals and slowly lose their portal inflow fare better than those whose portal flow is suddenly diverted at the time of total shunting. Further follow-up with angiography or flow studies, or both, may clarify this point. We have concluded from this trial that the distal splenorenal shunt is a technically more difficult
DISTAL
SPLENORENAL
SHUNT
TRIAL
429
operation than the portacaval shunt and is associated with a higher postoperative mortality rate in the learning phase. In experienced hands, it can be carried out with an operative mortality similar to portacaval shunt. It is as effective as portacaval shunt in preventing rebleeding. It is associated with a lower incidence of postoperative encephalopathy, particularly severe postshunt encephalopathy, and a better quality of life. However, improvement in longterm survival following this shunt as compared with portacaval shunt has not been demonstrated.
References 1. Warren
WD, Zeppa R, Fomon JJ. Selective trans-splenic decompression of gastroesophageal varices by distal splenorenal shunt. Ann Surg 1967;166:437-55, 2. Warren WD, Rudman D, Millikan W, et al. The metabolic basis of portasystemic encephalopathy and the effect of selective versus nonselective shunts. Ann Surg 1974;180:573-9. 3. Hutson DG, Pereiras R, Zeppa R, et al. The fate of esophageal varices following selective distal splenorenal shunt. Ann Surg 1976;183:496-501. 4. Mosimann R, Loup P. Efficacy and risks of the distal splenorenal shunt in the treatment of bleeding esophageal varices. Am J Surg 1977;133:163-8. 5. Silver DS, Puckett CL, McNeer JF, et al. Evaluation of selective transsplenic decompression of gastroesophageal varices. Am J Surg 1974;127:30-4. 6. Zeppa R, Hutson DG, Bergstresser PR, et al. Survival after distal splenorenal shunt. Surg Gynecol Obstet 1977;145:12-6. 7. Galambos JT, Warren WD, Rudman D. et al. Selective and total shunts in the treatment of bleeding varices. A randomized controlled trial. N Engl J Med 1976;295:1089-95. a. Reichle FA, Fahmy WF, Golsorkhi M. Prospective comparative clinical trial with distal splenorenal and mesocaval shunts. Am J Surg 1979;137:13-21. 9. Conn HO, Resnick RH, Grace ND, et al. Distal splenorenal shunt versus portal-systemic shunt: current status of a controlled trial. Hepatology 1981;1:151-60. 10. Zeppa R, Hensley GT, Levi JU, et al. The comparative survivals of alcoholics versus nonalcoholics after distal splenorenal shunt. Ann Surg 1978;187:510-4. 11. Langer B, Stone RM, Pate1 SC, Colapinto RF. A prospective randomized trial of the selective distal splenorenal shunt. Surg Gynecol Obstet 1980;150:45-8. 12. Widrich WC, Robbins AH, Nabseth DC, et al. Portal hypertension changes following selective splenorenal shunt surgery: evaluation by percutaneous transhepatic portal catheterization, venography, and cinefluorography. Radio 1976:121:
295-302. 13. Maillard JN, Flamant the distal
splenorenal
YM, Hay JM, Chandler JG. Selectivity shunt. Surgery 1979:86:663-71.
of