Transjugular intrahepatic portosystemic shunts: Preliminary results in 25 patients Jeanne M. LaBerge, MD, Ernest J. Ring, MD, John R. Lake, M D , Linda D. Ferrell, MD, Margaret M. Doherty, R N , R o y L. Gordon, MD, John P. Roberts, MD, Marc Y. Peltzer, MD, and Nancy L. Ascher, MD, PhD,
San Francisco, Calif. A retrospective study of transjugular intrahepatic shunts performed between June 1990 and June 1991 is reported. Twelve patients were actively bleeding at the time of the procedure; 12 other patients had had one to five bleeding episodes within the previous 2 weeks, and one patient had massive ascites from Budd-Chiari syndrome. Most patients had severe liver disease: 21 Child's class C, three Child's class B, and one Child's class A. Transjugular intrahepatic shunting was technically successful in all cases. Portal vein pressures were reduced on average from 36 - 7 cm H 2 0 to 22 - 6 cm H20. Variceal bleeding ceased after transjugular intrahepatic shunting in all patients who were actively bleeding. Five patients died (30-day mortality, 20%), and eight patients subsequently underwent elective liver transplantation. The transjugular intrahepatic shunts in the 12 other patients have remained patent an average of 5.5 months. Shunt occlusion occurred in three patients at 21, 24, and 102 days, respectively. All three occlusions were successfully reopened with percutaneous techniques, yielding a primary shunt patency of 88% and secondary shunt patency of 100%. Complications included new onset encephalopathy in one patient, which cleared with medical therapy and transient renal failure in one patient. These preliminary data suggest that transjugular intrahepatic shunting is a safe and effective therapy for the short-term treatment of patients with variceal hemorrhage, particularly in patients with severe liver disease awaiting transplantation. The long-term benefit of transjugular intrahepatic shunting awaits further follow-up. (J VASC SURG 1992;16:258-67.)
Variceal hemorrhage is a life-threatening consequence of portal hypertension for which several therapies are currently available, but each of the three common therapeutic alternatives has limitations in specific settings.1 Sclerotherapy successfully controls acute bleeding in most cases but frequently fails to prevent repeat bleeding. 2's Operative portosystemic shunting controls acute bleeding and reliably prevents rebleeding; however, the operative mortality rate in the emergency setting is high, and postoperative morbidity from hepatic failure or encephalopathy is unpredictable. 46 Liver transplantation controls bleeding and cures the underlying cause of portal hypertension, but transplantation is generally From the Department of Radiology (Drs. LaBerge, Ring, Gordon, Peltzer, and Ms. Doherty),the Departmentof Surgery (Drs. Roberts and Ascher), the Department of Medicine (Dr. Lake), and the Department of Pathology(Dr. Ferrell), University of California, San Francisco. Reprint requests: Jeanne M. LaBerge, MD, Department of Radiology, Box 0628, University of California, S.F., 505 Parnassus Ave., San Francisco, CA 94143-0628. 24/1/37161 258
impractical in the emergency setting, and it may be unnecessary in patients with mild liver disease once the variceal bleeding is controlled. 7,s Transjugular intrahepatic portosystemic shunting (TIPS) is a nonoperative therapeutic alternative for variceal hemorrhage. In this procedure, originally described by Rtsch et al.9 in 1969, portal decompression is achieved through a percutaneously created channel between a hepatic vein and the portal vein. Until recently, broad clinical application of TIPS was limited by a high incidence of early shunt occlusion. 10,11However, after the introduction of expandable metallic stents in the mid 1980s, several investigators demonstrated experimentally that prolonged shunt patency could be achieved by incorporating a stent within the shunt tract. 1214 By supporting the shunt wall, the stent prevented the elastic recoil of the liver parenchyma from occluding the shunt lumen. In 1986 Palmaz et al.~4 reported 100% shunt patency at 11 months in dogs with experimentally induced portal hypertension. So far, clinical experience with stented TIPS has
Volume 16 Number 2 August 1992
Transjugular intrahepatic portosystemic shunting: Preliminary r~ults 259
been small, and the role of this procedure in the management of variceal bleeding has not been fully determined. Other authors have recently published their results using the Palmaz Balloon-Expandable Stent (Johnson & Johnson, Warren, N.J.) to create a TIP shmat in a total of 17 patients. 15,16We have used the Wallstent (Schneider USA, Minneapolis, Minn.) self-expandable stent to construct a TIP shunt and have previously reported our early results in 13 patients awaiting liver transplantation. ~7 This report describes the overall results with TIPS in our first 25 consecutive patients. PATIENTS A N D M E T H O D S Patients Included in this retrospective study are the first 25 patients who underwent TIPS at the University of California, San Francisco (UCSF). These procedures were performed during a 1-year period from June 1990 to June 1991. This group consists of 14 men and 11 women, with a mean age of 45 years (range, 9 to 63 years). The indication for TIPS was variceal hemorrhage not responsive to sclerotherapy in 24 patients and Budd-Chiari syndrome with symptomatic ascites in one patient. Portal vein occlusion was a relative contraindication to the procedure. Twelve patients were actively bleeding at the time of the procedure. Twelve other patients had experienced at least one bleeding episode within 2 weeks before the procedure (range, 1 to 5). Supportive measures administered before TIPS included blood transfusion in 24 patients (average, 16 _+ 11 units; range, 2 to 40 units), intravenous vasopressin infusion in the 12 actively bleeding patients, and balloon tamponade in three patients. No specific therapy was given to correct any underlying coagulopathy or to reduce ascites before the procedure. Eighteen patients were considered candidates for transplantation at the time of the procedure. Esophageal or gastric varices were endoscopically identified before TIPS in all cases, and portal vein patency was assessed by preprocedural ultrasonography in aJl cases. Angiography was performed in five patients to further define the portal venous anatomy. The origin of the portal hypertension was hepatocellular disease in 23 patients, portal vein occlusion in one patient with Felty's syndrome, and hepatic vein occlusion in one patient with Budd-Chiari syndrome. Hepatoodlular disease was due to La~nnec's cirrhosis in 12 patients, posmecrotic cirrhosis in six patients, autoimrnune hepatitis in one patient, and cryptogenic cirrhosis in four patients.
The severity of liver disease as reflected by the Childs-Pugh modified scoring system was as follows: 21 Child's C, three Child's B, and one Child's A. Ascites was present in 22 patients (3 mild, 9 moderate, 10 severe), and clinically apparent encephalopathy was noted in nine patients. Mean preprocedural laboratory values included the following: hematocrit of 30% (range, 18 to 50), platelets of 106,000/M (range, 25,000 to 303,000), total serum bilirubin of 6.9 mg/dl (range, 0.4 to 22.7), serum albumin of 2.6 gm/dl (range, 1.9 to 3.6), and blood urea nitrogen of 23 mg/dl (range, 5 to 79). The mean prothrombin time was prolonged in 20 patients and was greater than 11/2 times normal in seven patients (mean, 15 seconds; range, 12.5 to 18.3 seconds). Four patients had undergone prior surgical portal decompression by a portacaval or mesocaval shunt. All four of these shunts were occluded at the time of the TIPS as demonstrated by arterial portography or portal venography. Twenty-four patients had undergone an average of 3.1 sclerotherapy sessions within 6 months before TIPS. Methods The TIPS protocol described herein was approved by the Committee on Human Research at the University of California, San Francisco. Informed consent was obtained from the patient or his or her next of kin. The procedure was performed in the radiology department, and patients were monitored in an intensive care unit for 24 hours after the procedure. A broad-spectrum antibiotic (usually 1 gm of ceftizoximine sodium) was administered before the procedure. Intravenous doses of sedative and analgesic medications were administered during the procedure as tolerated. The basic steps for performing TIPS are as follows: 1. A long, curved needle is introduced from the right internal jugular vein into the hepatic vein. 2. The needle is advanced through the fiver parenchyma into a branch of the portal vein. 3. The resultant hepatic vein to portal vein tract is catheterized and dilated. 4. Finally, an expandable metallic stent is deployed across the tract to create a permanent shunt. The metallic stent used in all of these procedures is the Wallstent endoprosthesis. This self-expanding wire mesh stent comes Ioaded onto a flexible 7F delivery catheter and is covered by a retractable plastic membrane. The stent is deployed by pulling back the plastic membrane allowing the wire mesh to expand and detach from the delivery catheter. These stents
260
LaBerge et aL
have a maximal diameter of 10 mm and are available in 4.2 and 6.8 cm lengths. A more detailed description of the TIPS technique is provided herein. Transjugular intrahepatic portosystemic shunting was performed in an angiographic room equipped with C-arm fluoroscopy, a tilting table, and digital subtraction angiography. In the first three patients, a 3F transhepatic catheter was initially placed in the portal vein to guide the transjugular portal vein puncture, but this guidance technique was subsequently considered unnecessary and was not used in any of the remaining cases. With the patient in the supine position, the right neck was cleansed with sterile solution, covered with a large transparent adhesive dressing, and isolated by sterile drapes. A right internal jugular venipuncture was performed, and an angiographic catheter and guide wire were advanced into the inferior vena cava. The catheter was exchanged for a 40 cm long 9F vascular sheath with hemostatic valve, and a curved angiographic catheter was manipulated into a the large right or middle hepatic vein. A sheathed Colapinto transjugular needle (Cook Inc., Bloomington, Ind.), composed of a 50 cm long, curved 16-gauge needle and an outer 45 cm long 9F catheter, was then advanced from the hepatic vein, in an anterior direction, 3 to 4 cm into the hepatic parenchyma toward the right portal vein. As suction was applied to the needle, it was slowly pulled back until blood was aspirated. Contrast was then injected to identify the vascular structure that had been punctured. After the portal vein was opacified, a flexible angiographic wire was advanced through the needle, manipulated into the main portal vein, and directed down either the superior mesenteric or splenic vein. The inner needle was then removed, and a 5F catheter was introduced over the wire through the 9F catheter. Initial portal pressures were measured, and a portal venogram was obtained. The 5F catheter was exchanged over a guide wire for an 8 mm angioplasty balloon catheter that was inflated across the parenchymal tract. The 9F transjugular needle sheath and angioplasty balloon were removed, and a 7F WaUstent catheter was advanced into the portal vein. The Wallstent was then deployed so that it fully covered the parenchymal tract and extended a short distance into both the hepatic vein and the portal vein. Either a 42 or a 68 mm long stent was used depending on the length of the tract. After stent deployment, the parenchymal segment of the stent was dilated with an 8 mm
Journal of VASCULAR SURGERY
angioplasty balloon catheter. Portal pressure measurements and venography were repeated. A second Wallstent was introduced, if necessary, to bridge any unstented segment of the shunt. If portal venography revealed persistent variceal flow, the Wallstent was further expanded with a 10 mm angioplasty balloon. When there was still hepatofugal flow into the varices despite dilation to 10 mm, the varices were selectively catheterized through the shunt and embolized with metallic coils. After completion of the shunt, final pressure measurements were obtained in the portal vein and inferior vena cava. The patient was then placed in a reverse Trendelenburg position. The catheter and sheath were removed, and the puncture site was compressed. Alternatively, in some patients at the completion of the procedure, the sheath was exchanged for a triple lumen central venous access catheter that was sutured to the patient's neck and maintained during the postprocedure period. Radiographs from a representative case illustrating the TIPS technique are provided in Fig. 1. Follow-up. All surviving patients have been followed by color-flow duplex Doppler ultrasound examinations of the liver performed before discharge and at 3-month intervals thereafter or until transplantation. Most patients rived in the local area and were monitored by regular clinic visits. Three patients were monitored by correspondence with referring physicians. Procedural and clinical data were obtained by a review of hospital and clinic records. Objective evidence of shunt patency was demonstrated by duplex Doppler examination or angiography. Clinical shunt patency was evidenced by the absence of variceal bleeding or worsening ascites. Histology. Histopathologic analysis of explant fiver specimens was performed in the patients who subsequently underwent orthotopic fiver transplantation. A cored-out segment of fiver containing the shunt was fixed in formalin and processed through paraffin. Transverse serial sections were obtained at 0.5 cm intervals and stained with hematoxylin and eosin, Masson's trichrome, and Verhoeffvan Gieson solution. Selected sections were also stained with Ulex europaeus immunoperoxidase. Slides were reviewed by fight microscopy, and luminal shunt diameters were measured with a measuring magnifier. RESULTS Procedural results Transjugular intrahepatic portosystemic shunting was successfully accomplished in all 25 patients. The procedure was performed in the standard fashion in
Volume 16 Number 2 August 1992
Transjugular intrahepatic portosystemic shunting: Preliminary results 261
Fig. 1. 36-year-old man with alcoholic liver disease had recurrent variceal bleeding despite prior mesocaval shtmt and live prior sclerotherapy sessions. TIPS tec~hnique used in this case illustrated above. A, Right hepatic vein -wasselectivelycatheterized and hepatic venography was performed. B, A tract was created from the right hepatic vein to a branch of the right portal vein. A 5F catheter was advanced into the main portal vein. C, A W~stent, deployed across the parenchymal tract, was dilated to 8 mm. Follow-up portal venogram demonstrates prompt flow through the shunt and no flow into varices.
24 patients. A modified technique (reported elsewhere) i8 was required in one patient whose anomalous hepatic venous anatomy precluded the formation of a shunt from a jugular approach; the shunt was successfully completed with use of a left femoral approach. In the patient with Budd-Chiari syndrome, TIPS was performed by creating a tract from the stump of the right hepatic vein to the right portal vein. Mean procedure time (calculated as the total time the patient spent in the radiology suite), available in 10 patients, was 2.2 hours (range, i to 3 hours). Portal pressure was reduced after TIPS from an initial average of 36.0 _+ 7 cm H 2 0 (range, 21 to 55) to 22.3 ___6 cm H 2 0 (range, 14 to 37). In all cases, the final postshurlt venogram demonstrated hepatopetal flow in the main portal vein and no flow into gastric or esophageal varices. In 12 cases, variceal flow was eliminated after dilation of the stent to 8 mm. In 10 of the remaining 13 patients, persistent variceal flow was eliminated by expanding the stent to 10 mm, and in three patients complete obliteration of variceal flow required concomitant n:anscatheter embolization. In eight cases, the final portal venogram demonstrated prograde flow in the intrahepatic branches of the portal vein as well as flow through the
shunt (Fig. 2,A and B). In the other 17 cases, there was total diversion of portal flow through the shunt and no demonstrable flow into the intrahepatic po~al branches (Fig. 3, A and B). Complications No immediate complications were noted during TIPS. Specifically, inadvertent puncture of the gallbladder, colon, or other viscus was not observed, and there were no bleeding or puncture site complications. Five patients died within 30 days of the procedure (30-day mortality, 20%). Mean time to death was 9 days. In all cases, the cause of death was acute respiratory distress syndrome, which was variously complicated by pneumonia, renal failure, and sepsis. Liver function did not deteriorate and variceal bleeding did not recur before death in any of these five patients. Of note, four deaths occurred in patients who were encephalopathic and vigorously bleeding during the procedure; the average transfusion requirement in these four patients was 26 units. The ~£th patient, who was neither encephalopathic nor actively bleeding, died of complications resulting from an esophageal perforation that occurred during previous sclerotherapy.
Journal of VASCULAR SURGERY
262 LaBerge et al.
Fig. 2. Partial portal diversion after TIPS. A, Portal venogram before TIPS. B, Portal venogram after completion of an 8 mm TIP shunt shows prograde flow into the right and left intrahepatic branches of the portal vein (arrows) as well as flow through the shunt.
Fig. 3. Total portal diversion after TIPS. A, Portal venogram before TIPS demonstrates filling of large esophageal varices (arrows). B, After completion of an 8 mm TIPS, flow to varices was criminated. Intrahepatic portal vein flow is now reversed; the direction of intrahepatic portal flow is now toward the TIPS.
Two of the patients who died also had spontaneous bacterial peritonitis before the procedure. Transjugular intrahepatic portosystemic shunting was performed in these two patients because they were actively bleeding (average transfusion requirement was 15 and 29 units, respectively) and because they were not considered to be viable operative
candidates (Childs C, Pugh score 14 and 15, respectively). Complications developed in two other patients within 30 days after TIPS. Transient renal failure with oliguria developed in one patient. His serum creatinine rose to a maximum of 7.1 mg/dl 1 week after TIPS and returned to a baseline of 2.1 mg/dl 2
Volume 16 Number 2 August 1992
Transjugular intrahepatic portosystemicshunting: Preliminary results 263
weeks later. Mild encephalopathy developed in i patient several weeks after TIPS, which cleared rapidly with medical management.
Follow-up All patients who were actively bleeding when the procedure began stopped bleeding immediately after TIPS. Doppler ultrasound examination performed on the day after the procedure confirmed shunt patency in all patients. Ascites resolved in all patients who survived the initial hospitalization. Mean total hospitalization time was 9.4 _ 5.7 days. Eight patients underwent elective orthotopic liver transplantation 4 to 106 days after TIPS (mean, 30 days). The indication for transplantation in these patients was end-stage liver disease, and the timing of transplantation was determined by patient preference and donor organ availability. The remaining 12 patients are alive and have not undergone liver transplantation. Mean follow-up in this group is 5.5 months (range, 1.4 to 10.9 months). Shunt occlusion was discovered in tl~:ee patients at 21, 24, and 102 days, respectively. A technical problem was responsible for shunt occlusion in one patient in whom recurrent ascites developed 9 days after TIPS. Reexamination of the angiogram obtained during initial stent placement revealed that a short parenchymal segment of the initial shunt had been left unstented. The occluded shunt was catheterized, and a new stent was deployed to bridge the previously unstented segment. The restented shunt has remained patent during a 7-month follow-up. Exuberant pseudointimal hyperplasia appears to have caused shunt occlusion in a second patient who bled again at 102 days. This patient underwent transplantation 4 ,days after the occluded shunt had been restented; histologic evaluation of the explant shunt specimen revealed that a layer of pseudointimal hyperplasia, 1200 Ixm thick had developed over the surface of the original stent. The cause of shunt occlusion in the third patient, a 9-year-old gift who bled again at 24 days, could not be determined. The occluded shunt was restented, and the patient underwent liver transplantation 16 days later. Histologic inspection of the shunt revealed no evidence of parenchymal in-growth through the original stent, and the pseudointima that had developed over the original stent was of only moderate thickness (maximum, 500 ~m). All three shunts were successfully reopened and restented with use of percutaneous techniques. Patency was maintained during the follow-up period in all of the restented shunts. None ,of the other 17 surviving patients devel-
Table I, Follow-up of surviving patients undergoing TIPS Patients Patient
Outcome
1 2 3 4 5 6 7 8 9 I0 11 12 13 14 15 16 17 18 19 20
T T T T T T T T NT NT NT NT NT NT NT NT NT NT NT NT
Clinical patency (days)
Confirmed* patency (days)
4 4 4 20 23 27 60 106 42 44 55 56 140 144 155 187 204
4 4 4 20 23 27 60t 106t 44 55 56 144# 155 187 167 170 322 205
277 322 326
Shunt occlusion (days)
24 102
21
T, Transplanted; NT, not transplanted. *By surgery or ultrasound. tSecondary patency.
oped symptoms suggestive of shunt occlusion during the follow-up period; that is, none bled again from varices or developed recurrent ascites. Shunt patency was documented at surgery in all patients undergoing transplantation. Shunt patency has been confirmed by follow-up ultrasound examination in 10 of i2 survivors who did not have transplants. Patency data for the 20 surviving patients are provided in Table I. On the basis of clinical follow-up, primary shunt patency is 88% and secondary shunt patency (calculated with Rutherford's criteria 19) is 100%.
Histology Gross examination of six explant liver specimens revealed that all shunts were patent and unchanged in size, shape, and position from initial placement. Histologic examination at 4 days revealed an irregular luminal surface with liver parenchyma protruding between the stent wires. A single, patchy layer of endothelial cells covered approximately 75% of the shunt surface. The three specimens obtained at 3 to 4 weeks all showed a smooth luminal shunt surface with a complete, contiguous single layer of endothelial cells lining the shunt surface. The stent wires were covered by a layer of granulation tissue that varied in thickness from 400 to 700 jxm. Representative histologic sections from a shunt sectioned at 3 weeks are shown in Fig. 4.
Journal of VASCULAR SURGERY
264 LaBerge et al.
Fig. 4. Histologic section. A, Histologic features of tissue at edge of stent stained by Ulex europeans (which highlights endothelial cells as black line)29 at day 2 1. Note endothelial staining of surface of pseudointima (top, black line). Underlying connective tissue contains holes where stent wires have been removed (larger white spaces not outlined by solid black line). Tissue surrounding the wire holes is fibrotic, with focal new vessel formation. Residual liver tissue is present below. (Ulex europeans immunoperoxidase stain; original magnification x 50.) B, Higher magnification of same area shows surface of pseudointima of stent and new vessel formation (endothelial surfaces outlined in black) in the underlying organizing connective tissue. Inflammatory cells are also present adjacent to wire holes and scattered throughout connective tissue. (Ulex europeaus immunoperoxidase stain; original magnification x 125; highlighting endothelial cells). Two specimens contained restented shunts. The first was restented at 102 days, and the patient had a transplant 4 days later. Histologic analysis of this shunt revealed a 1200 Ixm thick, heavily collagenized layer of scar tissue between the initial stent and the second stent. The second patient, a 9-year-old girl, had restenting at 3 weeks, and transplantation was performed 4 weeks later. Histologic evaluation revealed that a stratified pseudointima had formed
between the original stent and the shunt surface. The pseudointima was thickest in the parenchymal portion of the shunt where it measured 500 ~m. The layer ofpseudointima located between the two stents was composed of acellular, heavily collagenized scar tissue. The pseudointima superficial to the second stent was more cellular containing myofibroblastic proliferative cells and capillary ingrowth. Small areas of parenchymal hemorrhage (maxi-
Volume 16 Number 2 August 1992
Tmnsjugular intrahepatic portosystemic shunting: Preliminary results 265
m u m size, 2 x 5 mm) were present in the 4-day specimen, but appeared to have resolved ~uneventfially in the older specimens. Bile ducts and vessels in close proximity to the shunts were mildly compressed, but the parenchyma was otherwise unaffected. DISCUSSION
Although the data obtained in this retrospective study are preliminary, several conclusions are warranted. First, it is clear that percutaneous intrahepatic portosystemic shunts can be created safely and reliably with use of this technique. Second, TIPS is an effective means of lowering portal pressure and controlling variceal hemorrhage. Third, as previously reported, the procedure does not inhibit or impede subsequent liver transplantation.XS Finally, the risk of hepatic encephalopathy or liver failure is low. Conclusions regarding long-term shunt patency and postshunt morbidity require more extensive investigation, but our preliminary results suggest that early shunt occlusion can be expected in some patients and is easily remedied by percutaneous methods. Immediate complications from the TIPS procedure are unusual. We encountered no procedural complications in this study, and few complications have been noted at other centers. Stent migration occurred in one patient in the series reported by Zemel et al.,16 but this complication did not occur in our series. Because TIP shunts are constructed entirely within the liver, the risk of hemorrhage induced by the procedure is low. Any bleeding associated with the creation of a hepatic vein to portal vein tract flows into the shunt or the hepatic vein. Inadvertent puncture of the liver capsule could potentially lead to intraperitoneal hemorrhage, but this complication was not noted in any of our patients nor has it been reported. Similarly, inadvertent puncture of the gallbladder or colon is possible but could be avoided by cautious guidance of the transjugular puncture needle. Several hemodynamic features of the intrahepatic portosystemic shunt are worth noting. First, since changes in flow dynamics can be monitored during the procedure by repeat portal venography, and the shunt diameter can be adjusted by balloon dilation, the final size of the shunt can be tailored to each patient's specific flow dynamics. If the pressure remains markedly elevated and varices continue to fill after creation of an 8 m m shunt, the stent can easily be dilated to 10 ram. If there is stir persistent variceal flow, it can be eliminated, when necessary, by transcatheter embolization. Second, unfike the surgical portacaval shunts that divert flow into the
infrahepatic inferior vena cava, TIPS diverts flow into the suprahepatic inferior vena cava. This f~atute may result in more reliable flow diversion because shunt egress is maintained even when the intrahepatic cava is compressed by an enlarged fiver, and there is elevated pressure in the infrahepatic cava. Large caliber (16 to 22 mm) side-to-side portacaval shunts are associated with an unpredictable incidence of liver failure and hepatic encephalopathy. 22,2a Warren 24 and others have speculated that these complications are the result of hepatic portal flow deprivation caused by total portal diversion. Recently several authors have suggested that portacaval shunts that provide only partial portal &compression may result in a diminished incidence encephalopathy and liver failure. This contention is supported by data from both Sarfeh and Rypins and Sarfehz° and Johansen 21who reported an incidence of postshunt encephalopathy of 9% and 6%, respectively with smaJl stoma H-grafts. Average portal pressure reduction in our series was 14 cm H 2 0 , with an average residual pressure of 22 cm H20. Although our results are preliminary, encephalopathy has been noted in only one patient undergoing TIPS to date-yielding an incidence of 5% (in patients surviving longer than 1 week). This compares favorably with a 20% to 40%
San Francisco, Calif. A retrospective study of transjugular intrahepatic shunts performed between June 1990 and June 1991 is reported. Twelve patients were actively bleeding at the time of the procedure; 12 other patients had had one to five bleeding episodes within the previous 2 weeks, and one patient had massive ascites from Budd-Chiari syndrome. Most patients had severe liver disease: 21 Child's class C, three Child's class B, and one Child's class A. Transjugular intrahepatic shunting was technically successful in all cases. Portal vein pressures were reduced on average from 36 - 7 cm H 2 0 to 22 - 6 cm H20. Variceal bleeding ceased after transjugular intrahepatic shunting in all patients who were actively bleeding. Five patients died (30-day mortality, 20%), and eight patients subsequently underwent elective liver transplantation. The transjugular intrahepatic shunts in the 12 other patients have remained patent an average of 5.5 months. Shunt occlusion occurred in three patients at 21, 24, and 102 days, respectively. All three occlusions were successfully reopened with percutaneous techniques, yielding a primary shunt patency of 88% and secondary shunt patency of 100%. Complications included new onset encephalopathy in one patient, which cleared with medical therapy and transient renal failure in one patient. These preliminary data suggest that transjugular intrahepatic shunting is a safe and effective therapy for the short-term treatment of patients with variceal hemorrhage, particularly in patients with severe liver disease awaiting transplantation. The long-term benefit of transjugular intrahepatic shunting awaits further follow-up. (J VASC SURG 1992;16:258-67.)
Variceal hemorrhage is a life-threatening consequence of portal hypertension for which several therapies are currently available, but each of the three common therapeutic alternatives has limitations in specific settings.1 Sclerotherapy successfully controls acute bleeding in most cases but frequently fails to prevent repeat bleeding. 2's Operative portosystemic shunting controls acute bleeding and reliably prevents rebleeding; however, the operative mortality rate in the emergency setting is high, and postoperative morbidity from hepatic failure or encephalopathy is unpredictable. 46 Liver transplantation controls bleeding and cures the underlying cause of portal hypertension, but transplantation is generally From the Department of Radiology (Drs. LaBerge, Ring, Gordon, Peltzer, and Ms. Doherty),the Departmentof Surgery (Drs. Roberts and Ascher), the Department of Medicine (Dr. Lake), and the Department of Pathology(Dr. Ferrell), University of California, San Francisco. Reprint requests: Jeanne M. LaBerge, MD, Department of Radiology, Box 0628, University of California, S.F., 505 Parnassus Ave., San Francisco, CA 94143-0628. 24/1/37161 258
impractical in the emergency setting, and it may be unnecessary in patients with mild liver disease once the variceal bleeding is controlled. 7,s Transjugular intrahepatic portosystemic shunting (TIPS) is a nonoperative therapeutic alternative for variceal hemorrhage. In this procedure, originally described by Rtsch et al.9 in 1969, portal decompression is achieved through a percutaneously created channel between a hepatic vein and the portal vein. Until recently, broad clinical application of TIPS was limited by a high incidence of early shunt occlusion. 10,11However, after the introduction of expandable metallic stents in the mid 1980s, several investigators demonstrated experimentally that prolonged shunt patency could be achieved by incorporating a stent within the shunt tract. 1214 By supporting the shunt wall, the stent prevented the elastic recoil of the liver parenchyma from occluding the shunt lumen. In 1986 Palmaz et al.~4 reported 100% shunt patency at 11 months in dogs with experimentally induced portal hypertension. So far, clinical experience with stented TIPS has
Volume 16 Number 2 August 1992
Transjugular intrahepatic portosystemic shunting: Preliminary r~ults 259
been small, and the role of this procedure in the management of variceal bleeding has not been fully determined. Other authors have recently published their results using the Palmaz Balloon-Expandable Stent (Johnson & Johnson, Warren, N.J.) to create a TIP shmat in a total of 17 patients. 15,16We have used the Wallstent (Schneider USA, Minneapolis, Minn.) self-expandable stent to construct a TIP shunt and have previously reported our early results in 13 patients awaiting liver transplantation. ~7 This report describes the overall results with TIPS in our first 25 consecutive patients. PATIENTS A N D M E T H O D S Patients Included in this retrospective study are the first 25 patients who underwent TIPS at the University of California, San Francisco (UCSF). These procedures were performed during a 1-year period from June 1990 to June 1991. This group consists of 14 men and 11 women, with a mean age of 45 years (range, 9 to 63 years). The indication for TIPS was variceal hemorrhage not responsive to sclerotherapy in 24 patients and Budd-Chiari syndrome with symptomatic ascites in one patient. Portal vein occlusion was a relative contraindication to the procedure. Twelve patients were actively bleeding at the time of the procedure. Twelve other patients had experienced at least one bleeding episode within 2 weeks before the procedure (range, 1 to 5). Supportive measures administered before TIPS included blood transfusion in 24 patients (average, 16 _+ 11 units; range, 2 to 40 units), intravenous vasopressin infusion in the 12 actively bleeding patients, and balloon tamponade in three patients. No specific therapy was given to correct any underlying coagulopathy or to reduce ascites before the procedure. Eighteen patients were considered candidates for transplantation at the time of the procedure. Esophageal or gastric varices were endoscopically identified before TIPS in all cases, and portal vein patency was assessed by preprocedural ultrasonography in aJl cases. Angiography was performed in five patients to further define the portal venous anatomy. The origin of the portal hypertension was hepatocellular disease in 23 patients, portal vein occlusion in one patient with Felty's syndrome, and hepatic vein occlusion in one patient with Budd-Chiari syndrome. Hepatoodlular disease was due to La~nnec's cirrhosis in 12 patients, posmecrotic cirrhosis in six patients, autoimrnune hepatitis in one patient, and cryptogenic cirrhosis in four patients.
The severity of liver disease as reflected by the Childs-Pugh modified scoring system was as follows: 21 Child's C, three Child's B, and one Child's A. Ascites was present in 22 patients (3 mild, 9 moderate, 10 severe), and clinically apparent encephalopathy was noted in nine patients. Mean preprocedural laboratory values included the following: hematocrit of 30% (range, 18 to 50), platelets of 106,000/M (range, 25,000 to 303,000), total serum bilirubin of 6.9 mg/dl (range, 0.4 to 22.7), serum albumin of 2.6 gm/dl (range, 1.9 to 3.6), and blood urea nitrogen of 23 mg/dl (range, 5 to 79). The mean prothrombin time was prolonged in 20 patients and was greater than 11/2 times normal in seven patients (mean, 15 seconds; range, 12.5 to 18.3 seconds). Four patients had undergone prior surgical portal decompression by a portacaval or mesocaval shunt. All four of these shunts were occluded at the time of the TIPS as demonstrated by arterial portography or portal venography. Twenty-four patients had undergone an average of 3.1 sclerotherapy sessions within 6 months before TIPS. Methods The TIPS protocol described herein was approved by the Committee on Human Research at the University of California, San Francisco. Informed consent was obtained from the patient or his or her next of kin. The procedure was performed in the radiology department, and patients were monitored in an intensive care unit for 24 hours after the procedure. A broad-spectrum antibiotic (usually 1 gm of ceftizoximine sodium) was administered before the procedure. Intravenous doses of sedative and analgesic medications were administered during the procedure as tolerated. The basic steps for performing TIPS are as follows: 1. A long, curved needle is introduced from the right internal jugular vein into the hepatic vein. 2. The needle is advanced through the fiver parenchyma into a branch of the portal vein. 3. The resultant hepatic vein to portal vein tract is catheterized and dilated. 4. Finally, an expandable metallic stent is deployed across the tract to create a permanent shunt. The metallic stent used in all of these procedures is the Wallstent endoprosthesis. This self-expanding wire mesh stent comes Ioaded onto a flexible 7F delivery catheter and is covered by a retractable plastic membrane. The stent is deployed by pulling back the plastic membrane allowing the wire mesh to expand and detach from the delivery catheter. These stents
260
LaBerge et aL
have a maximal diameter of 10 mm and are available in 4.2 and 6.8 cm lengths. A more detailed description of the TIPS technique is provided herein. Transjugular intrahepatic portosystemic shunting was performed in an angiographic room equipped with C-arm fluoroscopy, a tilting table, and digital subtraction angiography. In the first three patients, a 3F transhepatic catheter was initially placed in the portal vein to guide the transjugular portal vein puncture, but this guidance technique was subsequently considered unnecessary and was not used in any of the remaining cases. With the patient in the supine position, the right neck was cleansed with sterile solution, covered with a large transparent adhesive dressing, and isolated by sterile drapes. A right internal jugular venipuncture was performed, and an angiographic catheter and guide wire were advanced into the inferior vena cava. The catheter was exchanged for a 40 cm long 9F vascular sheath with hemostatic valve, and a curved angiographic catheter was manipulated into a the large right or middle hepatic vein. A sheathed Colapinto transjugular needle (Cook Inc., Bloomington, Ind.), composed of a 50 cm long, curved 16-gauge needle and an outer 45 cm long 9F catheter, was then advanced from the hepatic vein, in an anterior direction, 3 to 4 cm into the hepatic parenchyma toward the right portal vein. As suction was applied to the needle, it was slowly pulled back until blood was aspirated. Contrast was then injected to identify the vascular structure that had been punctured. After the portal vein was opacified, a flexible angiographic wire was advanced through the needle, manipulated into the main portal vein, and directed down either the superior mesenteric or splenic vein. The inner needle was then removed, and a 5F catheter was introduced over the wire through the 9F catheter. Initial portal pressures were measured, and a portal venogram was obtained. The 5F catheter was exchanged over a guide wire for an 8 mm angioplasty balloon catheter that was inflated across the parenchymal tract. The 9F transjugular needle sheath and angioplasty balloon were removed, and a 7F WaUstent catheter was advanced into the portal vein. The Wallstent was then deployed so that it fully covered the parenchymal tract and extended a short distance into both the hepatic vein and the portal vein. Either a 42 or a 68 mm long stent was used depending on the length of the tract. After stent deployment, the parenchymal segment of the stent was dilated with an 8 mm
Journal of VASCULAR SURGERY
angioplasty balloon catheter. Portal pressure measurements and venography were repeated. A second Wallstent was introduced, if necessary, to bridge any unstented segment of the shunt. If portal venography revealed persistent variceal flow, the Wallstent was further expanded with a 10 mm angioplasty balloon. When there was still hepatofugal flow into the varices despite dilation to 10 mm, the varices were selectively catheterized through the shunt and embolized with metallic coils. After completion of the shunt, final pressure measurements were obtained in the portal vein and inferior vena cava. The patient was then placed in a reverse Trendelenburg position. The catheter and sheath were removed, and the puncture site was compressed. Alternatively, in some patients at the completion of the procedure, the sheath was exchanged for a triple lumen central venous access catheter that was sutured to the patient's neck and maintained during the postprocedure period. Radiographs from a representative case illustrating the TIPS technique are provided in Fig. 1. Follow-up. All surviving patients have been followed by color-flow duplex Doppler ultrasound examinations of the liver performed before discharge and at 3-month intervals thereafter or until transplantation. Most patients rived in the local area and were monitored by regular clinic visits. Three patients were monitored by correspondence with referring physicians. Procedural and clinical data were obtained by a review of hospital and clinic records. Objective evidence of shunt patency was demonstrated by duplex Doppler examination or angiography. Clinical shunt patency was evidenced by the absence of variceal bleeding or worsening ascites. Histology. Histopathologic analysis of explant fiver specimens was performed in the patients who subsequently underwent orthotopic fiver transplantation. A cored-out segment of fiver containing the shunt was fixed in formalin and processed through paraffin. Transverse serial sections were obtained at 0.5 cm intervals and stained with hematoxylin and eosin, Masson's trichrome, and Verhoeffvan Gieson solution. Selected sections were also stained with Ulex europaeus immunoperoxidase. Slides were reviewed by fight microscopy, and luminal shunt diameters were measured with a measuring magnifier. RESULTS Procedural results Transjugular intrahepatic portosystemic shunting was successfully accomplished in all 25 patients. The procedure was performed in the standard fashion in
Volume 16 Number 2 August 1992
Transjugular intrahepatic portosystemic shunting: Preliminary results 261
Fig. 1. 36-year-old man with alcoholic liver disease had recurrent variceal bleeding despite prior mesocaval shtmt and live prior sclerotherapy sessions. TIPS tec~hnique used in this case illustrated above. A, Right hepatic vein -wasselectivelycatheterized and hepatic venography was performed. B, A tract was created from the right hepatic vein to a branch of the right portal vein. A 5F catheter was advanced into the main portal vein. C, A W~stent, deployed across the parenchymal tract, was dilated to 8 mm. Follow-up portal venogram demonstrates prompt flow through the shunt and no flow into varices.
24 patients. A modified technique (reported elsewhere) i8 was required in one patient whose anomalous hepatic venous anatomy precluded the formation of a shunt from a jugular approach; the shunt was successfully completed with use of a left femoral approach. In the patient with Budd-Chiari syndrome, TIPS was performed by creating a tract from the stump of the right hepatic vein to the right portal vein. Mean procedure time (calculated as the total time the patient spent in the radiology suite), available in 10 patients, was 2.2 hours (range, i to 3 hours). Portal pressure was reduced after TIPS from an initial average of 36.0 _+ 7 cm H 2 0 (range, 21 to 55) to 22.3 ___6 cm H 2 0 (range, 14 to 37). In all cases, the final postshurlt venogram demonstrated hepatopetal flow in the main portal vein and no flow into gastric or esophageal varices. In 12 cases, variceal flow was eliminated after dilation of the stent to 8 mm. In 10 of the remaining 13 patients, persistent variceal flow was eliminated by expanding the stent to 10 mm, and in three patients complete obliteration of variceal flow required concomitant n:anscatheter embolization. In eight cases, the final portal venogram demonstrated prograde flow in the intrahepatic branches of the portal vein as well as flow through the
shunt (Fig. 2,A and B). In the other 17 cases, there was total diversion of portal flow through the shunt and no demonstrable flow into the intrahepatic po~al branches (Fig. 3, A and B). Complications No immediate complications were noted during TIPS. Specifically, inadvertent puncture of the gallbladder, colon, or other viscus was not observed, and there were no bleeding or puncture site complications. Five patients died within 30 days of the procedure (30-day mortality, 20%). Mean time to death was 9 days. In all cases, the cause of death was acute respiratory distress syndrome, which was variously complicated by pneumonia, renal failure, and sepsis. Liver function did not deteriorate and variceal bleeding did not recur before death in any of these five patients. Of note, four deaths occurred in patients who were encephalopathic and vigorously bleeding during the procedure; the average transfusion requirement in these four patients was 26 units. The ~£th patient, who was neither encephalopathic nor actively bleeding, died of complications resulting from an esophageal perforation that occurred during previous sclerotherapy.
Journal of VASCULAR SURGERY
262 LaBerge et al.
Fig. 2. Partial portal diversion after TIPS. A, Portal venogram before TIPS. B, Portal venogram after completion of an 8 mm TIP shunt shows prograde flow into the right and left intrahepatic branches of the portal vein (arrows) as well as flow through the shunt.
Fig. 3. Total portal diversion after TIPS. A, Portal venogram before TIPS demonstrates filling of large esophageal varices (arrows). B, After completion of an 8 mm TIPS, flow to varices was criminated. Intrahepatic portal vein flow is now reversed; the direction of intrahepatic portal flow is now toward the TIPS.
Two of the patients who died also had spontaneous bacterial peritonitis before the procedure. Transjugular intrahepatic portosystemic shunting was performed in these two patients because they were actively bleeding (average transfusion requirement was 15 and 29 units, respectively) and because they were not considered to be viable operative
candidates (Childs C, Pugh score 14 and 15, respectively). Complications developed in two other patients within 30 days after TIPS. Transient renal failure with oliguria developed in one patient. His serum creatinine rose to a maximum of 7.1 mg/dl 1 week after TIPS and returned to a baseline of 2.1 mg/dl 2
Volume 16 Number 2 August 1992
Transjugular intrahepatic portosystemicshunting: Preliminary results 263
weeks later. Mild encephalopathy developed in i patient several weeks after TIPS, which cleared rapidly with medical management.
Follow-up All patients who were actively bleeding when the procedure began stopped bleeding immediately after TIPS. Doppler ultrasound examination performed on the day after the procedure confirmed shunt patency in all patients. Ascites resolved in all patients who survived the initial hospitalization. Mean total hospitalization time was 9.4 _ 5.7 days. Eight patients underwent elective orthotopic liver transplantation 4 to 106 days after TIPS (mean, 30 days). The indication for transplantation in these patients was end-stage liver disease, and the timing of transplantation was determined by patient preference and donor organ availability. The remaining 12 patients are alive and have not undergone liver transplantation. Mean follow-up in this group is 5.5 months (range, 1.4 to 10.9 months). Shunt occlusion was discovered in tl~:ee patients at 21, 24, and 102 days, respectively. A technical problem was responsible for shunt occlusion in one patient in whom recurrent ascites developed 9 days after TIPS. Reexamination of the angiogram obtained during initial stent placement revealed that a short parenchymal segment of the initial shunt had been left unstented. The occluded shunt was catheterized, and a new stent was deployed to bridge the previously unstented segment. The restented shunt has remained patent during a 7-month follow-up. Exuberant pseudointimal hyperplasia appears to have caused shunt occlusion in a second patient who bled again at 102 days. This patient underwent transplantation 4 ,days after the occluded shunt had been restented; histologic evaluation of the explant shunt specimen revealed that a layer of pseudointimal hyperplasia, 1200 Ixm thick had developed over the surface of the original stent. The cause of shunt occlusion in the third patient, a 9-year-old gift who bled again at 24 days, could not be determined. The occluded shunt was restented, and the patient underwent liver transplantation 16 days later. Histologic inspection of the shunt revealed no evidence of parenchymal in-growth through the original stent, and the pseudointima that had developed over the original stent was of only moderate thickness (maximum, 500 ~m). All three shunts were successfully reopened and restented with use of percutaneous techniques. Patency was maintained during the follow-up period in all of the restented shunts. None ,of the other 17 surviving patients devel-
Table I, Follow-up of surviving patients undergoing TIPS Patients Patient
Outcome
1 2 3 4 5 6 7 8 9 I0 11 12 13 14 15 16 17 18 19 20
T T T T T T T T NT NT NT NT NT NT NT NT NT NT NT NT
Clinical patency (days)
Confirmed* patency (days)
4 4 4 20 23 27 60 106 42 44 55 56 140 144 155 187 204
4 4 4 20 23 27 60t 106t 44 55 56 144# 155 187 167 170 322 205
277 322 326
Shunt occlusion (days)
24 102
21
T, Transplanted; NT, not transplanted. *By surgery or ultrasound. tSecondary patency.
oped symptoms suggestive of shunt occlusion during the follow-up period; that is, none bled again from varices or developed recurrent ascites. Shunt patency was documented at surgery in all patients undergoing transplantation. Shunt patency has been confirmed by follow-up ultrasound examination in 10 of i2 survivors who did not have transplants. Patency data for the 20 surviving patients are provided in Table I. On the basis of clinical follow-up, primary shunt patency is 88% and secondary shunt patency (calculated with Rutherford's criteria 19) is 100%.
Histology Gross examination of six explant liver specimens revealed that all shunts were patent and unchanged in size, shape, and position from initial placement. Histologic examination at 4 days revealed an irregular luminal surface with liver parenchyma protruding between the stent wires. A single, patchy layer of endothelial cells covered approximately 75% of the shunt surface. The three specimens obtained at 3 to 4 weeks all showed a smooth luminal shunt surface with a complete, contiguous single layer of endothelial cells lining the shunt surface. The stent wires were covered by a layer of granulation tissue that varied in thickness from 400 to 700 jxm. Representative histologic sections from a shunt sectioned at 3 weeks are shown in Fig. 4.
Journal of VASCULAR SURGERY
264 LaBerge et al.
Fig. 4. Histologic section. A, Histologic features of tissue at edge of stent stained by Ulex europeans (which highlights endothelial cells as black line)29 at day 2 1. Note endothelial staining of surface of pseudointima (top, black line). Underlying connective tissue contains holes where stent wires have been removed (larger white spaces not outlined by solid black line). Tissue surrounding the wire holes is fibrotic, with focal new vessel formation. Residual liver tissue is present below. (Ulex europeans immunoperoxidase stain; original magnification x 50.) B, Higher magnification of same area shows surface of pseudointima of stent and new vessel formation (endothelial surfaces outlined in black) in the underlying organizing connective tissue. Inflammatory cells are also present adjacent to wire holes and scattered throughout connective tissue. (Ulex europeaus immunoperoxidase stain; original magnification x 125; highlighting endothelial cells). Two specimens contained restented shunts. The first was restented at 102 days, and the patient had a transplant 4 days later. Histologic analysis of this shunt revealed a 1200 Ixm thick, heavily collagenized layer of scar tissue between the initial stent and the second stent. The second patient, a 9-year-old girl, had restenting at 3 weeks, and transplantation was performed 4 weeks later. Histologic evaluation revealed that a stratified pseudointima had formed
between the original stent and the shunt surface. The pseudointima was thickest in the parenchymal portion of the shunt where it measured 500 ~m. The layer ofpseudointima located between the two stents was composed of acellular, heavily collagenized scar tissue. The pseudointima superficial to the second stent was more cellular containing myofibroblastic proliferative cells and capillary ingrowth. Small areas of parenchymal hemorrhage (maxi-
Volume 16 Number 2 August 1992
Tmnsjugular intrahepatic portosystemic shunting: Preliminary results 265
m u m size, 2 x 5 mm) were present in the 4-day specimen, but appeared to have resolved ~uneventfially in the older specimens. Bile ducts and vessels in close proximity to the shunts were mildly compressed, but the parenchyma was otherwise unaffected. DISCUSSION
Although the data obtained in this retrospective study are preliminary, several conclusions are warranted. First, it is clear that percutaneous intrahepatic portosystemic shunts can be created safely and reliably with use of this technique. Second, TIPS is an effective means of lowering portal pressure and controlling variceal hemorrhage. Third, as previously reported, the procedure does not inhibit or impede subsequent liver transplantation.XS Finally, the risk of hepatic encephalopathy or liver failure is low. Conclusions regarding long-term shunt patency and postshunt morbidity require more extensive investigation, but our preliminary results suggest that early shunt occlusion can be expected in some patients and is easily remedied by percutaneous methods. Immediate complications from the TIPS procedure are unusual. We encountered no procedural complications in this study, and few complications have been noted at other centers. Stent migration occurred in one patient in the series reported by Zemel et al.,16 but this complication did not occur in our series. Because TIP shunts are constructed entirely within the liver, the risk of hemorrhage induced by the procedure is low. Any bleeding associated with the creation of a hepatic vein to portal vein tract flows into the shunt or the hepatic vein. Inadvertent puncture of the liver capsule could potentially lead to intraperitoneal hemorrhage, but this complication was not noted in any of our patients nor has it been reported. Similarly, inadvertent puncture of the gallbladder or colon is possible but could be avoided by cautious guidance of the transjugular puncture needle. Several hemodynamic features of the intrahepatic portosystemic shunt are worth noting. First, since changes in flow dynamics can be monitored during the procedure by repeat portal venography, and the shunt diameter can be adjusted by balloon dilation, the final size of the shunt can be tailored to each patient's specific flow dynamics. If the pressure remains markedly elevated and varices continue to fill after creation of an 8 m m shunt, the stent can easily be dilated to 10 ram. If there is stir persistent variceal flow, it can be eliminated, when necessary, by transcatheter embolization. Second, unfike the surgical portacaval shunts that divert flow into the
infrahepatic inferior vena cava, TIPS diverts flow into the suprahepatic inferior vena cava. This f~atute may result in more reliable flow diversion because shunt egress is maintained even when the intrahepatic cava is compressed by an enlarged fiver, and there is elevated pressure in the infrahepatic cava. Large caliber (16 to 22 mm) side-to-side portacaval shunts are associated with an unpredictable incidence of liver failure and hepatic encephalopathy. 22,2a Warren 24 and others have speculated that these complications are the result of hepatic portal flow deprivation caused by total portal diversion. Recently several authors have suggested that portacaval shunts that provide only partial portal &compression may result in a diminished incidence encephalopathy and liver failure. This contention is supported by data from both Sarfeh and Rypins and Sarfehz° and Johansen 21who reported an incidence of postshunt encephalopathy of 9% and 6%, respectively with smaJl stoma H-grafts. Average portal pressure reduction in our series was 14 cm H 2 0 , with an average residual pressure of 22 cm H20. Although our results are preliminary, encephalopathy has been noted in only one patient undergoing TIPS to date-yielding an incidence of 5% (in patients surviving longer than 1 week). This compares favorably with a 20% to 40%
