Bureau C, Garcia-Pagan JC, Otal P, Pomier-Layrargues G, Chabbert V, Cortez C, et al. Improved Clinical Outcome Using Polytetrafluoroethylene-Coated Stents for TIPS: Results of a Randomized Study. Gastroenterology 2004:126;469–475. (Reprinted with permission from The American Gastroenterological Association.)
This study compares transjugular intrahepatic portosystemic shunts (TIPS) using expanded-polytetrafluroethylene (ePTFE) covered-stent grafts with those using conventional bare stents. The results are so much in favor of the covered stent group that previous randomized studies using bare stents may become irrelevant, and new studies with covered stents are required. The study is in accordance with previous studies on patency and outcome after TIPS using the ePTFE covered stent graft.1–4
The high patency of the covered stent (1- and 2-year probability of 86% and 80%, respectively) is the key to the improved outcome variables. Only 5 of the 39 patients developed shunt dysfunction, which was defined as a reduction of the lumen of the shunt at angiography of >50% or a pressure gradient of >12 mm Hg. Three of the 5 patients who had a pressure gradient of >12 mm Hg did not show any narrowing of the shunt lumen at angiography. Unfortunately, details such as the gradients obtained at the end of the TIPS-implantation and at revision were not given. Assuming that the gradients were 11 or 12 mm Hg after TIPS implantation, these shunts were, per definition, regarded as dysfunctional when the gradient increased by only 1 to 2 mm Hg. Such a minute change may be due to a type 2 error in the pressure measurement or caused by a slight increase in hepatic or decrease in mesenteric resistance. It cannot be denoted as shunt insufficiency as long as angiography or duplex-sonography is unequivocally negative. In this regard, the definition of shunt dysfunction by the threshold of 12 mm Hg may be criticized. It would better be replaced by a relative change—for example, a 25% loss of the gain achieved by the TIPS. Such a definition, also used in coronary stenting, may possibly reduce shunt dysfunction to 2 patients (5%) who showed greater elevations in the pressure gradients together with the respective radiological finding of shunt dysfunction.
As a consequence of the higher patency rate, a clinical relapse (rebleeding or ascites) was seen in only 7.7% in the covered-stent group, compared to 29.3% in the bare-stent group. Survival was higher in the covered-stent group, but the difference was not statistically significant. However, a significant improvement in survival has recently been shown in a retrospective study comparing covered TIPS with bare-stent TIPS.4 Despite the better patency and the higher efficacy of the covered stent, it was accompanied not with a higher but in fact with an even lower rate of hepatic encephalopathy (HE) (1-year probability of 22% vs. 41%, respectively). The authors explain this finding by the decreased need to revise the TIPS in the covered-stent group.
The groups were comparable in all but one variable, the shunt diameter, which was significantly greater in the bare-stent group (11.7 ± 0.8 mm vs. 10.5 ± 0.9 mm). The difference of 1.2 mm is not negligible, since it results in a difference in the stent area or shunt volume of 500 mL/min with a respective additional reduction in the pressure gradient by about 20%. Strangely, the authors did not discuss this finding. They also do not give any information about how the given diameters were assessed. Were they the means of the nominal stent diameters or of the balloons used for their expansion, or were they assessed sonographically? The latter mode may be the most accurate, since it best reflects the true size. Independent from the mode of measurement it is unlikely that the covering of the stent will have caused a systematic error. Why, then, are the diameters different? The pressure gradients of the groups before TIPS were identical (20 ± 7 mm Hg for covered stent and 20 ± 5 mm Hg for bare stent) and indicate no reason to implant different stent sizes. It can be assumed that the investigators chose greater stents in the bare-stent group to prevent early insufficiency and smaller stents in the covered-stent group to reduce the risk of HE. In this regard it is surprising that the post-TIPS pressure gradients were also identical between the groups (7 ± 4 mm Hg).
The difference in the shunt diameter may have biased the result. On the one hand, the smaller diameter of the covered stent was sufficient to prevent rebleeding. On the other hand, the greater diameter of the bare stents increased shunt-related side effects such as liver failure and encephalopathy. Accordingly, the higher rate of deaths in the bare-stent group (19 vs. 12) was mainly due to liver failure (5 vs. 1). Like loss of liver function, shunt-induced HE is also closely related to the shunt diameter.5 After implantation of bare stents, HE occurs predominantly early after the intervention, at a time when the shunt has its maximum diameter. This fact is also demonstrated in this study where about 2 out of 3 of the HE episodes occurred within the first month of follow-up. Due to intimal proliferation, the bare-stent lumen reduces with time, rapidly approaching the smaller size of the covered stent. As a result, the slopes of the probability curves of patients' remaining free of encephalopathy parallel one another after the first month of follow-up. The same will be true after TIPS revision. Taking into account prior experience of HE after de novo stent placement should help in choosing smaller-diameter stents when TIPS revision seems warranted.
The favorable results of this study imply that bare stents may not be used anymore. This implication is, however, not justified. Their higher rate of narrowing may also be a blessing. First, shunt-induced liver failure and HE cannot be predicted with sufficient accuracy. Even if one limits the TIPS treatment to good candidates (age < 65 years, bilirubin < 3 mg/dL, no previous episodes of HE), liver failure and HE can still occur. In these patients, the dynamic process of size reduction is welcome, and in a good number of patients, self-adjustment of the optimal compromise between shunt function and shunt side effects may occur unless disturbed by unjustified shunt revision. Therefore, shunt revision should be performed only if a clinical relapse is impending or existing. Second, in patients who have a chance of improvement of the liver disease, a transient reduction in the portal pressure may be favorable, and the permanent shunting would probably be a disadvantage. This positive development, which may be seen in about 20% of alcoholic patients with cirrhosis as well as in a subgroup of patients with Budd-Chiari syndrome cannot be predicted. Third, shunt revision is, in most cases, an easy, fast, and safe procedure comparable with the measurement of the wedged hepatic venous pressure. Its major disadvantage is the need for permanent control of shunt function and estimation of the risk of relapse in variceal bleeders, which may increase costs.6
If both types of stents are useful, a guideline for the decision of which stent should be used for which patient is needed. The protection against a relapse has high priority in variceal bleeders, because they may rapidly die from rebleeding, but not in ascites patients. Therefore, bleeders need a safe prophylaxis to reduce mortality from rebleeding. Since most rebleedings can be prevented by a partial pressure reduction by 25% to 50%,7–10 small-diameter (8-mm) covered shunts may be implanted to achieve the best ratio between benefits and side effects. Only the covered stent, which does not narrow with time by proliferation of the neointima, opens up the possibility of inserting custom-made, thin-lumen shunts from the start. Thus, instead of the bare stent with a diameter of 10–12 mm in use hitherto, a bare stent with an internal diameter of 8 mm may now become the norm. The high patency rate together with a smaller stent diameter may reduce both the risk of rebleeding and the incidence of HE. Thus, the use of large-diameter shunts (most with a diameter of 12 mm) in this study by Bureau et al. cannot be recommended. In contrast to variceal rebleeding, sudden death is unlikely if ascites relapses. Therefore, shunt control is not necessary as long as the patient is free of ascites. Accordingly, there is no rationale that survival will be improved by the implantation of a Viatorr stent. In addition, patients with refractory ascites have not only a higher risk of HE but also a greater need for a large-diameter shunt (10–12 mm) to achieve a good response. In these patients the bare stent offers good responses early after implantation and protects against HE at long-term follow-up because of its natural narrowing. Instead, the implantation of a covered stent with a diameter large enough to reduce the gradient below 12 mm Hg has a lifelong, very high risk of inducing HE, which is clearly of greater relevance than the higher risk of relapse after bare-stent implantation.
In summary, this study is important because it shows that the covered Viatorr stent dramatically improves shunt patency and significantly reduces clinical relapse. As with surgical shunts, scheduled shunt control may no longer be necessary. The finding of a lower risk of HE and a trend toward improved survival is biased by the greater shunt diameter in the bare-stent group. It is assumed that bare stents may have a lower risk of HE when comparable diameters are applied. Covered stents offer the chance of small-diameter shunts in patients with variceal bleeding where partial pressure reduction may be effective in the prevention of rebleeding and may reduce HE. Bare stents may still be used in patients with a higher risk of HE and liver failure and in patients with refractory ascites. In these patients, the natural reduction in the lumen of the bare stent is often desired. If the shunt is tolerated, a covered stent may be implanted at revision.