Distal Splenorenal Shunt Versus Transjugular Intrahepatic Portal Systematic Shunt for Variceal Bleeding: A Randomized Trial

GASTROENTEROLOGY 2006;130:1643–1651

Distal Splenorenal Shunt Versus Transjugular Intrahepatic Portal Systematic Shunt for Variceal Bleeding: A Randomized Trial J. MICHAEL HENDERSON,* THOMAS D. BOYER,‡ MICHAEL H. KUTNER,§,储 JOHN R. GALLOWAY,储 LAYTON F. RIKKERS,¶ LENNOX J. JEFFERS,# KAREEM ABU–ELMAGD,** JASON CONNOR,§,‡‡ and the DIVERT Study Group *Department of Surgery, Cleveland Clinic, Cleveland, Ohio; ‡Department of Medicine, University of Arizona, Tuscon, Arizona; §Data Coordinating Center, Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, Ohio; 储Department of Surgery and School of Public Health, Emory University, Atlanta, Georgia; ¶Department of Surgery, University of Wisconsin, Madison, Wisconsin; #Department of Medicine, University of Miami, Miami, Florida; **Department of Surgery, University of Pittsburgh, Pittsburgh, Pennsylvania; and ‡‡Department of Statistics and H. John Heinz III School of Public Health Policy, Carnegie Mellon University, Pittsburgh, Pennsylvania

Background & Aims: Variceal bleeding refractory to medical treatment with ␤-blockers and endoscopic therapy can be managed by variceal decompression with either surgical shunts or transjugular intrahepatic portal systemic shunts (TIPS). This prospective randomized trial tested the hypothesis that patients receiving distal splenorenal shunts (DSRS) would have significantly lower rebleeding and encephalopathy rates than TIPS in management of refractory variceal bleeding. Methods: A prospective randomized controlled clinical trial at 5 centers was conducted. One hundred forty patients with Child–Pugh class A and B cirrhosis and refractory variceal bleeding were randomized to DSRS or TIPS. Protocol and event follow-up for 2– 8 years (mean, 46 ⴞ 26 months) for primary end points of variceal bleeding and encephalopathy and secondary end points of death, ascites, thrombosis and stenosis, liver function, need for transplant, quality of life, and cost were evaluated. Results: There was no significant difference in rebleeding (DSRS, 5.5%; TIPS, 10.5%; P ⴝ .29) or first encephalopathy event (DSRS, 50%; TIPS, 50%). Survival at 2 and 5 years (DSRS, 81% and 62%; TIPS, 88% and 61%, respectively) were not significantly different (P ⴝ .87). Thrombosis, stenosis, and reintervention rates (DSRS, 11%; TIPS, 82%) were significantly (P < .001) higher in the TIPS group. Ascites, need for transplant, quality of life, and costs were not significantly different. Conclusions: DSRS and TIPS are similarly efficacious in the control of refractory variceal bleeding in Child–Pugh class A and B patients. Reintervention is significantly greater for TIPS compared with DSRS. Because both procedures have equivalent outcomes, the choice is dependent on available expertise and ability to monitor the shunt and reintervene when needed.

ortal hypertension, the most common complication of cirrhosis, can lead to bleeding varices, ascites, hepatorenal syndrome, and hepatic encephalopathy.

P

Esophageal varices develop in half of the patients with cirrhosis and bleed in approximately 15%–20%. The risk of bleeding in patients with large varices is approximately 30% over 2 years.1 Once the varices have bled, the risk of rebleeding without intervention is ⬃70% over 2 years.2– 4 Seventy to eighty percent of patients who have bled at least once from varices can be treated effectively by a noncardioselective ␤-blocker and band ligation.5,6 However, 20%–30% of patients will fail such treatment and become candidates for variceal decompression with a surgical or a radiologic shunt.6 Decompression of varices can be achieved with surgical shunts, and numerous randomized studies comparing different surgical shunts in patients who have bled have been published. All types of surgical shunts prevented rebleeding in ⬎90% of patients, with no clear advantage of one type of surgical shunt on survival.5–7 Surgical shunts have been shown to prevent more effectively rebleeding than endoscopic therapy but without an improvement in survival.8 Current data on the selective distal splenorenal shunt (DSRS) in patients with good liver function (Child–Pugh class [CPC] A/B patients) show a hospital mortality of ⬍5%, rebleeding rates of 5%– 8%, and 3–5 year survival in the 75%– 80% range. DSRS has been regarded by many as the best therapy to prevent rebleeding in patients who have failed medical therapy yet have well-preserved liver function.9 –12 TIPS became widely available in the 1990s and is currently the most commonly used method of decompressing the portal venous system.13 A series of randomAbbreviations used in this paper: CPC, Child–Pugh class; DCC, data coordinating center; DSRS, distal splenorenal shunt; TIPS, transjugular intrahepatic portal systemic shunt. © 2006 by the American Gastroenterological Association Institute 0016-5085/06/$32.00 doi:10.1053/j.gastro.2006.02.008

1644

HENDERSON ET AL

ized trials have compared TIPS with endoscopic sclerotherapy14 and pharmacologic therapy15 and have documented better control of bleeding and no change in mortality but a higher encephalopathy rate and cost in the TIPS as compared with the medical therapy groups.16 The rebleeding rates after TIPS in these trials averaged 20%. It remains undefined whether TIPS or a surgical shunt is best for patients who have failed medical therapy. One trial has compared TIPS with a surgical 8-mm portocaval H graft interposition shunt.17 This study entered “all-comers” and included 46% CPC C patients and showed a benefit to surgical shunt compared with TIPS. However, the patients were not truly randomized but were entered into the study sequentially. The above-mentioned uncertainty about which type of shunt was best led to this prospective randomized controlled trial, which was initiated in 1996. The specific aims of this study were to compare DSRS and TIPS for (1) prevention of variceal rebleeding and (2) rates of encephalopathy in CPC A and B patients with bleeding refractory to ␤-blockers and/or endoscopic therapy. We hypothesized that TIPS patients would have higher rates of both. This trial ran from 1996 to 2004, and 140 patients were enrolled at 5 clinical centers. The last patient was entered in February 2002, and the median follow-up is 45 months (range, 0 –93 months). This is the first report of this trial.

Materials and Methods Organizational Structure The 5 clinical centers were The Cleveland Clinic Foundation, Emory University, University of Miami, University of Wisconsin, and the University of Pittsburgh. The data coordinating center (DCC) was directed through Quantitative Health Sciences at the Cleveland Clinic Foundation. The Principal Investigator and DCC personnel visited the clinical centers prior to initiating the study, and the DCC performed a data audit at each site. The DCC monitored follow-up, with 86% of scheduled visits completed and another 9% with telephone call follow-up. Review committees for deaths, clinical events, and radiology data received patient deidentified data from the DCC for consensus decision on defined end points. The Data Safety Monitoring Board met annually to review data and conduct of the trial and conducted planned interim analyses after 70 and 100 patients were enrolled to look at efficacy and futility as well as serious adverse events as stopping criteria.

Study Design The study design is presented as a flow diagram in Figure 1. Patient eligibility criteria were as follows: (1) endoscopically proven variceal bleeding secondary to cirrhosis of

GASTROENTEROLOGY Vol. 130, No. 6

any etiology; (2) liver disease severity CPC A or B18; and (3) failed adequate endoscopic therapy (sclerosis or banding), defined as (a) a single bleed sufficient to cause hypotension, (b) a 3-unit transfusion requirement with a bleeding episode, (c) multiple portal hypertensive bleeding episodes that had not required transfusion, or (d) not suitable for endoscopic or pharmacologic therapy. Exclusion criteria were as follows: patients with variceal bleeding with an etiology other than cirrhosis, bleeding from portal hypertensive gastropathy, a prior shunt procedure, medically intractable ascites, renal insufficiency with a creatinine ⬎2, age ⬍18 years, portal vein thrombosis on imaging study, polycystic liver disease, CPC score ⬎9, or compromized right heart function as defined clinically.

Patient Evaluation, Procedures, and Follow-up Potential patients were evaluated prior to randomization with the following: (1) demographics of age, sex, race, and institution; (2) confirmation of variceal bleeding by history and endoscopy; (3) assessment of liver disease by history, physical examination, and laboratory data, CPC, and (after 2000) MELD19 scores were calculated and hepatitis serologies obtained; (4) vascular imaging with ultrasound and angiography was done to determine suitability for either procedure.20,21 Randomization occurred when patients completed evaluation, signed informed consent, and entry criteria were validated by the DCC. Patients were randomized by center with a variable 2 or 4 permuted block size design, stratifying by CPC A and B, and alcoholic and nonalcoholic patients. Randomization was overseen by the DCC. The number of patients entered, the stratifications by center, and the prerandomization details are given in Table 1. All shunt procedures were done within 5 days of randomization. DSRS was performed by a total of 14 surgeons at the 5 centers, who all had significant experience with this technique, and the method varied little and followed previously published standards.22 TIPS was performed by 20 radiologists, who all had extensive experience, and there was a designated senior radiologist at each institution. TIPS were created through a right transjugular approach with the transhepatic tract dilated to 8 –12 mm and Wallstents (Boston Scientific, Natick, MA) used in all cases.13,23 Covered stents were not used as primary therapy in this study. The diameter of the stent was expanded to reduce the portal vein to right atrial gradient to ⬍12 mm Hg. If there was still filling of varices via the left gastric vein on portal injection of contrast, these varices could be embolized. Shunt patency was documented prior to discharge by ultrasound examination at 24 hours. Follow-up was either an elective protocol visit or an event-driven evaluation. Protocol visits were planned at 2, 6, and 12 weeks; at 6 and 12 months; and every 6 months thereafter. All visits included a clinical assessment with evaluation for encephalopathy and ascites and laboratory data. Protocol planned ultrasound evaluation was performed

May 2006

DSRS VS TIPS FOR REFRACTORY VARICEAL BLEEDING

1645

Figure 1. Patient flowchart.

for both groups at 6 months, 1 year, and annually thereafter; protocol shunt catheterizations were done for both groups at 1 year and annually in the TIPS group. When reintervention was required for thrombosis or stenosis, with shunt dilation or additional stent, further ultrasound studies at 6 and 12 weeks and 6 months were initiated for monitoring. Clinical event follow-up occurred when events happened. These included both inpatient and outpatient care at different times from protocol visits. The guidelines for these event “follow-up” visits were to provide optimal patient care and to submit all data related to that event to the DCC. Variceal bleeding, encephalopathy, and ascites events were reviewed and adjudicated by a Clinical Endpoint Review Committee, which was blinded to the study site. The study protocol was reviewed and approved by institutional review boards for each participating clinical center. The study and the protocol were explained to the patients and their

families by the PIs and study coordinators at each center preceding signing of the consent.

End Points The primary end points in this trial were variceal rebleeding and hepatic encephalopathy. Variceal rebleeding was defined as a fall in hematocrit of 6 points or a 2-unit transfusion requirement to restore the hematocrit to a preevent level. The event had to be associated with hematemesis or melena and required endoscopic confirmation with findings compatible with bleeding from varices or portal hypertensive gastropathy. Hepatic encephalopathy was primarily defined at a clinical level with neuropsychologic dysfunction sufficient to require hospitalization and/or a significant change in diet with protein restriction and/or medication with the addition of lactulose by the managing physician. These primary end points were reviewed by the Clinical Endpoint Review Committee,

1646

HENDERSON ET AL

GASTROENTEROLOGY Vol. 130, No. 6

Table 1. Baseline Patient Characteristics by Treatment Assignment DSRS

TIPS

Prior bleeding 86% 88% No. episodes 2.9 ⫾ 2.0 2.7 ⫾ 2.6 No. units 7.5 ⫾ 7.5 7.3 ⫾ 6.8 Pharmacotherapy 70% 66% Endotherapy 70% 75% Time from last bleed 42 ⫾ 56 52 ⫾ 76 (days) Varices at randomization (%) Size I/II/III/IV 27/24/36/13 14/45/24/17 Red color signs 50 63 Isolated gastric varices 38 39 Gastropathy 29/43 44/26 (mild/moderate) Prior encephalopathy (%) 19 13 Precipitated/spontaneous 14/4 13/0 Prior ascites (%) 52 52 Transient/persistent 51/1 50/0 At randomization: Age (mean ⫾ SD), y 53 ⫾ 10 52 ⫾ 1 Male/female 42/31 44/23 Child–Pugh A/B 41/32 39/28 CPC score (range) 6.4 ⫾ 1.1 (5–9) 6.3 ⫾ 1.0 (5–8) Alcoholic/nonalcoholic 43/30 37/30 Drinking in prior month 33% 16% Center: 1 18 16 2 18 17 3 15 13 4 13 13 5 9 8 MELD score 9.7 ⫾ 2.4 9.9 ⫾ 2.3 Lab data Total bilirubin (mg/dL) 1.3 ⫾ 0.7 1.2 ⫾ 0.6 Albumin (g/dL) 3.1 ⫾ 0.6 3.3 ⫾ 0.6 INR (ratio) 1.2 ⫾ 0.2 1.3 ⫾ 0.2 Serum creatinine (mg/dL) 0.8 ⫾ 0.2 0.8 ⫾ 0.2 Platelets (1000/mL) 92.3 ⫾ 42.3 109.5 ⫾ 53.7 AST (U/L) 61.6 ⫾ 36.5 50.7 ⫾ 34.5

which received summary data on all events in a blinded manner from the DCC. The secondary end points reported are death; ascites; shunt stenosis and thrombosis; and need for liver transplantation, liver tests, quality of life, and cost. Death was considered for any cause; was reviewed by a death review committee; and was categorized whether it was procedure related, liver disease related, or not associated with either. Shunt thrombosis was defined as total occlusion of the shunt; if suspected on Doppler ultrasound for TIPS or DSRS, this required angiographic confirmation. Shunt stenosis was not defined for either TIPS or DSRS on ultrasound evaluation but, if suspected by Doppler velocity flow changes, was defined by angiographic shunt catheterization and the following criteria for either shunt: (1) an absolute gradient ⬎15 mm Hg, (2) an increase of 5 mm Hg over postprocedure gradient and an absolute gradient ⱖ12 mm Hg, or (3) a ⬎50% stenosis with films reviewed by the Radiology Review Committee. Reintervention was only allowed or undertaken unless these criteria were documented.

Ascites was measured clinically and on ultrasound examination, was reviewed by the Clinical Endpoint Review Committee, and was judged an end point if therapy was changed. Quality of life was measured with the SF 36 prior to randomization and annually.24 This tool was chosen because there were no validated liver disease specific quality-of-life instruments available at the initiation of this trial. Cost analysis was performed using Diagnosis Related Group (DRG)based costs for inpatient events and Current Processing Terminology procedures (CPT)-based costs for outpatient events. All events for all patients from initial intervention and including all follow-up events were captured at all centers.

Sample Size Calculation Sample size was calculated from the literature data available in 1995 for the primary end points of variceal bleeding and encephalopathy following DSRS and TIPS. Data at that time showed TIPS rebleeding rates at 23% and incidence of new encephalopathy of 34%, with DSRS rebleeding rates at 5.4% and incidence of new encephalopathy of 14%. Taking into account the higher proportion of CPC C patients in the published TIPS trials, we estimated a 15% incidence difference for these 2 primary end points. Sample size estimates for a 2-sided test with power value of 0.90 and a 0.05 type I error gave a sample size of 139.25

Data Management and Analysis Data management was conducted using Oracle database (Oracle, Redwood Shores, CA) and data analyses performed using SAS (SAS Institute, Cary, NC). All analyses were performed as intent to treat.26 Study samples are summarized with count and percentages for categoric variables and means (⫾SD) or medians for continuous variables. Primary outcomes were compared by time to first clinically significant event using log rank tests and plotted using Kaplan–Meier curves. Subset analyses were performed comparing CPC A vs B patients and alcoholic vs nonalcoholic patients. Comparisons of proportions between TIPS vs DSRS and between centers were performed using ␹2 tests. Comparisons of continuous variables were performed using either Student t test or its nonparametric equivalent the Wilcoxon rank sum test. All reported P values are 2-sided; P values of .05 or less were considered statistically significant.

Results Patient Population Table 1 summarizes the patient characteristics at study entry by treatment assignment including bleeding history and management, variceal characteristics, and past and present liver status. There are no significant differences between the 2 groups for any of these parameters. Centers entered equivalent numbers of patients, given that center 5 was added after year 2.

May 2006

DSRS VS TIPS FOR REFRACTORY VARICEAL BLEEDING

Table 2. Summary of Number of Deaths by Center and Whether Procedure and/or Liver Disease Related

By center 1 2 3 4 5 Total 30-day mortality Procedure-related: Total Operative mortality Sepsis Progressive liver failure Renal failure Variceal bleed Related to liver disease: Total Progressive liver failure Hepatocellular carcinoma Sepsis Other Not related to liver disease: Total Cardiopulmonary Trauma/suicides Cancers Sepsis Unknown

DSRS

TIPS

8 5 8 5 2 28 5 6 2 2 1 1 0 15 8 4 2 1 7 4 0 1 1 1

2 10 5 7 3 27 1 3 0 1 1 0 1 15 11 3 0 1 9 1 4 1 1 2

1647

Outcomes Figure 1 summarizes details of the procedure hospitalization, length of follow-up, and number of patients in each group reaching primary and secondary end points. Further outcomes data are given in Table 2 and Figure 2. Each of these will be presented in more detail. Procedural complications. During the initial hospitalization, postprocedural complications documented a higher incidence of ascites in the DSRS group (10% vs 0%, respectively; P ⬍ .05); a higher rate of reintervation in the TIPS group (11% vs 1%, respectively; P ⬍ .05); and no significant differences in bleeding (0% both groups), encephalopathy (4% both groups), infection (14% DSRS, 8% TIPS), or death (5.6% DSRS, 1.5% TIPS). Rebleeding rates. Rebleeding rates were not significantly different at 5.5% in the DSRS and 10.5% in the TIPS group (P ⫽ .29; Figures 1 and 2). Subset analysis by CPC score or etiology showed no significant differences. The time to rebleeding in the DSRS patients was 1 week and 1, 2, and 65 months; whereas the time to rebleeding in the TIPS group was 1 week and 4, 5, 6, 34, 57, and 87 months. Encephalopathy. The time to the first encephalopathy event showed no significant difference in the 2

Figure 2. Kaplan–Meier curves for the main end points in the DIVERT Trial. Variceal rebleeding, first hepatic encephalopathy episode, and survival show no significant difference at any timepoint to 5 years. The reintervention rate was significantly (P ⬍ .001) greater in the TIPS compared with the DSRS group at all time points.

1648

HENDERSON ET AL

GASTROENTEROLOGY Vol. 130, No. 6

Table 3. Summary of SF36 Quality-of-Life Scores for Standardized Physical and Mental Components at Baseline and 1 year DSRS

Physical componenta Baseline 1 Year Mental componenta Baseline 1 Year aNormal bValues

TIPS

N

Median (25th/75th)b

N

Median (25th/75th)b

67 55

37.8 (32.7/45.0) 41.7 (33.4/50.5)

64 55

38.4 (28.6/45.7) 42.1 (36.1/49.3)

67 55

40.6 (32.8/50.7) 44.3 (33.6/55.2)

64 55

45.2 (35.6/52.6) 44.2 (37.3/56.0)

scores for physical and mental components for SF36 are 50. A lower score indicates poorer QOL. in parentheses are percentiles.

groups (Figure 2). By 5 years, 50% of patients in each group had at least 1 event of encephalopathy: half of these patients (25% of each group) had more than 1 episode of encephalopathy by 5 years. Subset analysis by CPC score or etiology showed no significant difference between the 2 groups. Survival. There was no significant survival difference between DSRS and TIPS patients (P ⫽ .87; Figure 2). The 2 and 5 year survivals in the TIPS group were 88% and 61% and after DSRS were 81% and 62%, respectively. Subset analysis showed no significant difference in mortality by CPC score or etiology between groups. There are no significant differences in procedural related, liver disease related, or other causes of death between the DSRS and TIPS groups (Table 2). Reintervention, stenosis, and thrombosis. Reintervention, stenosis, and thrombosis rates were significantly higher (P ⬍ .001) in the TIPS group compared with the DSRS group (Figure 2). Fifty-five patients (82%) in the TIPS group required dilation to maintain patency, whereas only 8 patients (11%) in the DSRS group required reintervention (P ⬍ .001). Significantly more patients in the TIPS group (15 patients) had total thrombosis compared with 2 DSRS patients (P ⬍ .01). The DSRS thromboses occurred at ⬍1 and 12 months,

whereas, in the TIPS group, 8 thromboses occurred at ⬍1 month and 7 at later follow-up (mean, 23 ⫾ 20 months). Ascites. There was no overall difference in incidence between DSRS and TIPS, but the pattern was different. Eleven of 21 patients in the DSRS group who developed ascites had a single episode within 2 months of operation. In the other 10, ascites occurred with endstage liver disease. All of the ascites episodes after TIPS occurred at least 2 months after the procedure, and 4 of the 17 patients in this group who developed ascites died within 3 months of developing their ascites. Transplantation. Transplantation was performed in 6 DSRS and 8 TIPS patients. The need for transplantation was similar across the 5 clinical centers. Two patients have died following transplantation, and the others continued to be followed to the end of the study with respect to mortality. Quality of life. Quality of life was measured with the SF 36 and showed no significant differences between the 2 groups at 1 year. The standardized physical and mental components at study entry and at 1 year are given in Table 3 and, at 1 year, are almost identical. Cost analysis data are summarized in Table 4. Cost analysis data show no overall significant difference in the

Table 4. Cost Analysis up to 5 Years in DSRS and TIPS Patients DSRS

TIPS

Time

No. patients

Cost

No. patients

Cost

Initial procedure 1y Alive Died 3y Alive Died 5y Alive Died

73

$28,734

67

$21,607

64 9

$36,886 $42,418

62 5

$35,691 $52,505

.42 .95

52 21

$46,664 $52,167

55 12

$54,084 $58,293

.32 .43

47 26

$48,796 $54,975

44 23

$70,527 $74,267

.18 .03

NOTE. Results are expressed as median cost based on DRG (inpatient) and CPT (outpatient) coding of all events at all centers.

P value

May 2006

cost of managing patients with either procedure. The initial cost of the surgical shunt is higher. On this DRG CPT-based cost analysis, the 1- and 3-year costs for both alive and dead patients are remarkably similar. At 5 years, there is a definite trend for a higher cost in TIPS patients, with cost being significantly higher in the subset of TIPS patients who died compared with the equivalent DSRS cohort. However, this must be viewed in light of a trend for earlier death in the DSRS group.

Discussion This trial was designed to evaluate the relative efficacy of surgical vs radiologic decompression of varices for bleeding that was refractory to medical treatment. The primary end point of greatest clinical importance in this trial is variceal rebleeding. The 5.5% rebleeding rate after DSRS is similar to that reported by others.9 –12 The use of DSRS in the community to manage this type of patient is likely to achieve similar results as reported in the literature over the past decade. The rebleeding rate of 10.5% after TIPS in this trial is the lowest reported in the literature and was achieved with meticulous follow-up and a high reintervention rate (82%). Our rebleeding rate is lower than the 13% rate reported using covered stents in the recent multicenter trial of covered vs uncovered TIPS stents.27 The reported reintervention rate (13%) in that trial at 300 days indicates that a lower reintervention rate can probably be achieved with covered TIPS but requires validation.27 However, it is only with intensive monitoring of either bare or coated stents that similar low rates of rebleeding with TIPS can be achieved. Standard protocols for follow-up of these patients following creation of a TIPS similar to those used in this study need to be developed when TIPS is used in the community. Survival was excellent in both groups in this trial. Procedural mortality was 6.9% and 1.5% for the DSRS and TIPS group, respectively, and was similar to published data for CPC A/B patients.9 –15 The overall 2- and 5-year survival of 80% and 63%, respectively, support the use of decompressive shunt for good-risk patients with refractory variceal bleeding. Such patients do not need transplantation, and the survival curves in Figure 2 look very similar to many published survival curves of patients undergoing liver transplantation.28,29 Although ours is a different patient population from those receiving liver transplantation, the current findings reemphasize the value of decompression of varices in the overall treatment repertoire for patients with CPC A and B cirrhosis and refractory variceal bleeding. This study also shows that, as we improve our ability to control variceal

DSRS VS TIPS FOR REFRACTORY VARICEAL BLEEDING

1649

bleeding, other causes of mortality start to emerge in this population, such as hepatoma. However, there was no difference in deaths from hepatoma in the 2 groups. Encephalopathy and ascites are clinical markers of decompensated liver disease and showed no significant difference in the 2 groups. In these CPC A and B patients, at least 1 episode of clinical encephalopathy occurred in 50% of patients by 5 years. The 1-year rate of 25% fits with other published TIPS data.13,14 The pattern of encephalopathy is remarkably similar between the 2 groups in terms of single vs multiple episodes and the occurrence of encephalopathy as a harbinger of death or need for transplantation. Figure 2 shows the higher early occurrences of encephalopathy in the first year in the TIPS group, but any difference disappears thereafter. This may relate to early diversion of portal flow in the TIPS group. However, the major risk factor for encephalopathy in both groups appears to be progression of the liver disease rather than portal diversion in CPC A and B patients. The ascites pattern is somewhat different in the 2 groups in that the DSRS patients had a peak occurrence of ascites early after operation. This fits with previously reported data, and the 10% incidence of ascites in the first month after DSRS puts a definitive incidence rate on this complication, which is otherwise poorly documented in the literature. Somewhat surprisingly, there is a high rate of ascites in the TIPS group at late follow-up, and, for patients in whom this occurred, it was a marker of advanced liver disease. Quality of life and cost are important secondary outcomes of this trial, with the initial data presented in this paper showing no significant difference in the 2 groups. More sophisticated analysis combining cost/quality of life/survival is beyond the scope of this paper. The data presented show the increasing cost of TIPS over time driven by the reintervention rate, but overall events lead to similar overall cost for the 2 groups. This study shows that for the primary end points of variceal rebleeding and hepatic encephalopathy, DSRS and TIPS are of similar efficacy for the prevention of rebleeding in patients who have failed medical therapy. There is no significant difference in survival, with excellent outcomes out to 5 years for both groups. A high reintervention rate is necessary to maintain TIPS patency and achieve the low rebleeding rate attained in this trial. The decision as to which therapy should be used for a given patient should be based on local expertise and ability of the patient to receive close follow-up care should a TIPS be the chosen therapy. In patients who have limited access to health care, a DSRS is clearly the therapy that requires the least observation. Irrespective of

1650

HENDERSON ET AL

the therapy selected, patients will require monitoring for the development of hepatocellular carcinoma. After the first year, if ascites or hepatic encephalopathy occur, then referral to a transplantation center is appropriate because these are indicative of advanced liver disease.

Appendix DIVERT Study Group Clinical Centers Cleveland Clinic: J. Michael Henderson David S. Barnes Michael Geisinger Mark Sands Kathy Agnor Susan Curtas Emory University: Thomas D. Boyer John R. Galloway Enrique Martinez Qiang Cai Louis G. Martin Joanne Halcomb Sharon Behan Alicia Whitcher University of Miami: Lennox Jeffers Enrique G. Molina Arie Regev Joe U. Levi Alan S. Livingstone Jorge Guerra Jr Jeffrey Raskin K. Rajender Reddy Eugene R. Schiff Andreas Tzakis Jose Yrizarry Edward Russell Thomas Scagnelli Joy Ness Adrianna Arcila Amy Siebert University of Pittsburgh: Kareem Abu-Elmagd Kapil B. Chopra Albert Zajko John J. Fung Geoffrey Bond F. Leland Thaete C. Andrew Bonham Hugo Vargas

GASTROENTEROLOGY Vol. 130, No. 6

Diane Morton Darlene Koritsky Holly Chapman Cindy Blaha University of Wisconsin: Layton F. Rikkers Kevin Block (died 2001) Michael Lucey Munci Kalayoglu Myron Wojtowycz Ian Sproat Deb Chicks Janice Cooney Data Coordinating Center Michael Kutner Jason Connor Marlene Goormastic Susana Arrigain Adrienne Heerey Vicky Konig Debbie Bick Consultants University of Arizona: Thomas D. Boyer Data Safety Monitoring Board Douglas Labrecque-Iowa (Chair) Donald Jensen-Chicago Theodore Karrison-Chicago Jeanne LaBerge-San Francisco Roger Jenkins-Boston Patricia Robuck-NIDDK-Project Officer

References 1. North Italian Endoscopic Club (NIEC) for the study and treatment of esophageal varices. Prevention of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter trial. N Engl J Med 1988;319:983–989. 2. Garcia-Tsao G. Current management of the complications of cirrhosis and portal hypertension: variceal bleeding, ascites and spontaneous bacterial peritonitis. Gastroenterology 2001;120: 726 –748. 3. D’Amico G, DeFranchis R, and a Co-operative Study Group. Upper digestive bleeding in cirrhosis. Post-therapeutic outcome and prognostic indicators. Hepatology 2003;38:599 – 612. 4. Sharara AI, Rockey DC. Gastroesophageal variceal hemorrhage. N Engl J Med 2001;345:669 – 681. 5. D’Amico G, Pagliaro L, Bosch J. The treatment of portal hypertension: a meta-analytic review. Hepatology 1995;22:332–353. 6. Boyer TD, Henderson JM. Portal hypertension and bleeding esophageal varices. In: Zakim D, Boyer TD, eds. Hepatology: a textbook of liver disease. 4th ed. Philadelphia: W.B. Saunders, 2002;581– 629. 7. Henderson JM. Variceal bleeding: which shunt? Gastroenterology 1986;91:1021–1023. 8. Spina GP, Henderson JM, Rikkers LF, Teres J, Burroughs AK, Conn HO, Pagliaro L, Santambrogio R. Distal spleno-renal shunt versus endoscopic sclerotherapy in the prevention of variceal

May 2006

9.

10. 11.

12.

13.

14.

15.

16.

17.

18.

19.

rebleeding: a metaanalysis of 4 randomized controlled trials. J Hepatol 1992;16:338 –345. Henderson JM, Nagle A, Curtas S, Geisinger M, Barnes D. Surgical shunts and TIPS for variceal decompression in the 1990s. Surgery 2000;128:540 –547. Rikkers LF. The changing spectrum of treatment for variceal bleeding. Ann Surg 1998;228:536 –546. Jenkins RL, Gedaly R, Pomposelli JJ, Pomfret EA, Gordon F, Lewis WD. Distal splenorenal shunt: role, indications, and utility in the era of liver transplantation. Arch Surg 1999;134:416 – 420. Orozco H, Mercado MA. The evolution of portal hypertension surgery: lessons from 1000 operations and 50 years’ experience. Arch Surg 2000;135:1389 –1393. Boyer TD, Haskal ZJ. The role of transjugular intrahepatic portosystemic shunt in the management of portal hypertension. Hepatology 2005;41:386 – 400. Papatheodoridis GV, Goulis J, Leandro G, Patch D, Burroughs AK. Transjugular intrahepatic portosystemic shunt compared with endoscopic treatment for prevention of variceal rebleeding: a metaanalysis. Hepatology 1999;30:612– 622. Escorsell AM, Banares R, Garcia-Pagan JC, Gilabert R, Moitinho E, Piqueras B, et al. TIPS versus drug therapy in preventing variceal rebleeding in advanced cirrhosis: a randomized controlled trial. Hepatology 2002;35:385–392. Meddi P, Merli M, Lionetti R, DeSantis A, Valeriano V, Masini A, et al. Cost analysis for the prevention of variceal rebleeding: a comparison between transjugular intrahepatic portosystemic shunt and endoscopic sclerotherapy in a selected group of Italian cirrhotic patients. Hepatology 1999;29:1074 –1077. Rosemurgy AS, Serafini FM, Zweibel BR, Black TJ, Kudryk BT, Nord HR, et al. Transjugular intrahepatic portosystemic shunt vs. small-diameter prosthetic H-graft portacaval shunt: extended follow-up of an expanded randomized prospective trial. J Gastroinest Surg 2000;4:589 –597. Pugh RN, Murray Lyan IM, Dawson JL, Pietroni MC, Williams R. Transection of the esophagus for bleeding esophageal varices. Br J Surg 1973;60:646 – 649. Malinchoc M, Kamath PS, Gordon FD, Peine CJ, Rank J, ter Borg PC. A model to predict poor survival in patients undergoing transjugular intrahepatic portosystemic shunts. Hepatology 2000;31: 864 – 871.

DSRS VS TIPS FOR REFRACTORY VARICEAL BLEEDING

1651

20. Ozaki CF, Anderson JC, Rikkers LF. Duplex ultrasound as a noninvasive technique of assessing portal hemodynamics. Am J Surg 1988;155:70. 21. Nordlinger BM, Nordlinger DF, Fulenwider JT, et al. Angiography in portal hypertension: clinical significance in surgery. Am J Surg 1980;139:132. 22. Henderson JM, Rikkers LF. Atlas of liver surgery— distal splenorenal shunt. In: Bell RH, Rikkers LF, Mulholland MW, eds. Digestive tract surgery. Philadelphia: Lippincoff-Raven, 1996:781–788. 23. Haskal ZJ, Martin L, Cardella JF, Cole P, Drooz A, Grassi CJ, et al. Quality improvement guidelines for transjugular intrahepatic portosystemic shunts. J Vasc Interv Radiol 2001;12:131–136. 24. Ware JE, Snow KK, Kosinski M, Gandak B. SF-36 Health survey: manual and interpretation. Boston: The Health Institute, New England Medical Center, 1993. 25. Shuster JJ. Handbook of sample size guidelines for clinical trials. London, UK: CRC Press Inc, 1990. 26. Fisher LD, Dixon DO, Herson J, Frankowski RK, Hearron MS, Peace KE. 1990. Intention to treat in clinical trials. In: Peace KE, ed. Statistical issues in drug research and development. Philadelphia: Marcel Dekker, 1990:331–350. 27. Bureau C, Garcia-Pagan JC, Otal P, Pomier-Layrargues G, Chabbert V, Cortez C, Perreault P, Peron JM, Abraldes JG, Bilbao JI, Bosch J, Rousseau H, Vinel JP. Improved clinical outcome using polytetrafluoroethylene-coated stents for TIPS: results of a randomized study. Gastroenterology 200;126:469 – 475. 28. Henderson JM, Hooks M, Galloway JR, Dodson TF, Boyer T, Gilmore GT, Kutner MH, Hood MM. Selective shunt in the management of variceal bleeding in the era of liver transplantation. Ann Surg 1992;216:248 –255. 29. Rikkers LF, Jin G, Langnas AN, Shaw BW Jr. Shunt surgery during the era of liver transplantation. Ann Surg 1997;226:51–57. Received November 17, 2005. Accepted January 25, 2006. Address requests for reprints to: J. Michael Henderson, MD, Division of Surgery, E-32, Cleveland Clinic, 9500 Euclid Avenue, Cleveland, Ohio 44195. e-mail: [email protected]; fax: (216) 444 8510. Supported by the NIH through NIDDK: DK050680 and, for the study group, by GCRCs from participating sites: Cleveland, M01 RR 018390; Emory, M01 00039; Miami, M01 RR 16587; Wisconsin, M01 96-74001; Pittsburgh M01 RR 00056. Members of the DIVERT Study Group are listed in Appendix.