OBJECTIVE. The purpose of this article is to describe the evaluation of transjugular intrahepatic portosystemic shunts (TIPS) and the management of dysfunctional shunts.
CONCLUSION. TIPS can become dysfunctional if stenosis develops in the shunt or the hepatic vein above the shunt. Screening allows detection of stenoses before portal hypertensive–related complications recur. Revision of stenotic shunts can be easily accomplished in most cases. Techniques for screening and revision will be discussed.
Although transjugular intrahepatic portosystemic shunts (TIPS) have become integral to the management of portal hypertensive–related complications, stenosis of the shunt has been a major problem. Originally, when TIPS were all created with bare metal stents, the loss of primary patency was around 50% at 1 year after shunt creation. The introduction of polytetrafluoroethylene (PTFE)–covered stents has vastly improved patency but stenosis still occurs in 8–20% of patients at 1 year after TIPS creation [1–5].
The significance of TIPS stenoses is that they can lead to recurrent portal hypertension and put patients at risk for reaccumulation of ascites or further variceal hemorrhage. It is important for interventional radiologists to actively follow their patients with TIPS to assess how well the shunt is functioning. The interventional radiologists understand the shunt better and should be able to make better assessments regarding the function of the TIPS and the need for revision. Furthermore, if you place the shunt and have others follow the patient, then your role as physician is diminished to that of technician.
Evaluating TIPS Function
There is no universal agreement on protocol for when or how often to screen TIPS function. One approach is to simply wait for symptoms of portal hypertension to recur. This approach can be used in patients whose TIPS were placed to treat ascites. This approach is not advisable for patients with a history of variceal bleeding because the first symptom suggesting a problem with the TIPS might be a fatal recurrent hemorrhage. Even in patients with ascites, detecting shunt problems before the patient becomes especially symptomatic is beneficial because TIPS venography is much easier when the patient does not have a large volume of ascites. The patient is more comfortable and able to breathe easier in a supine position when there is not much ascites. Furthermore, massive ascites forces the liver more cephalad, and the extra density caused by the ascites degrades the fluoroscopic image; both of these effects make the procedure more difficult.
The timing of screening has also not been standardized and varies greatly between institutions. Our protocol has been to perform Doppler ultrasound at 1, 3, 6, and 12 months after TIPS creation and every 6–12 months thereafter, depending on the patient’s clinical status. Evaluations at other time intervals may be triggered by any recurrence of ascites or bleeding. Although this schedule has not been scientifically validated, it has served us well in our relatively large TIPS experience (1223 patients since 1991) and is similar to protocols used by others. Through use of this protocol, many hemodynamically significant TIPS stenoses have been detected and fixed before the patients experienced recurrent hemorrhage. In recent years, the improved patency of TIPS made possible by the development of a dedicated PTFE-covered device (Viatorr, W. L. Gore) has allowed some loosening of this schedule and even caused some authors  to speculate that routine screening may not be necessary for TIPS created with Viatorr stents. Again, the ideal timing of follow-up screening in this era of stent-grafts has not been scientifically determined.
Ultrasound has been the primary tool used to screen for TIPS stenoses because it is noninvasive, readily available, and relatively low cost compared with other imaging modalities. However, many different ultrasound parameters have been used to assess patency, with variable results.
Flow velocities in the TIPS represent the primary parameter. A single velocity measurement in the mid shunt has been used by some, although the sensitivity and specificity with this methodology were only 86% and 54%, respectively, in one study . However, with careful examination of velocities along the length of the shunt, it is often possible to identify a specific stenosis with a major change in velocities across the stenosis. Whereas some authors [7–9] use a drop in shunt velocities to below 40–60 cm/s as the criterion for calling a stenosis, it is also possible to measure significantly elevated velocities (over 200 cm/s) in a jet-effect zone just beyond the stenosis (Figs. 1A, 1B, 1C, 1D, 1E, 1F, 1G, and 1H).
However, if the stenosis is very close to the inferior vena cava (IVC) within the hepatic vein, the jet effect will not be seen and the velocities will be uniformly low within the TIPS (Figs. 2A, 2B, 2C, and 2D). Thus, velocities in the shunt should range between 90 and 190 cm/s in most patent TIPS, and peak velocities below or above this range may indicate a stenosis .
Main portal vein velocity is another useful parameter. Our group previously showed that before TIPS the main portal velocity is usually 20 cm/s but after TIPS it typically increases to more than 30 cm/s . When a shunt gets stenotic, the flow in the portal vein leading up to the TIPS is diminished and the main portal velocity drops often down below 30 cm/s. Others have used higher values, such as 40 cm/s , but they have also reported ultrasound to be an inaccurate screening tool. The direction of flow in the portal vein branches should also be evaluated. In most patients who have pre-TIPS hepatopetal flow (toward the liver), the flow direction reverses and becomes hepatofugal after creation of the TIPS. When a stenosis develops, the flow in these branches often reverts to hepatofugal .
We have routinely used ultrasound with a high degree of confidence to screen TIPS function. In a study that compared ultrasound criteria to venographic proof of stenosis, it was found that no individual parameter was more than 84% specific in predicting TIPS dysfunction. However, when an overall assessment was made by considering all the parameters, the sensitivity and specificity for detecting TIPS stenoses were 92% and 72%, respectively . Also in that study, it was shown that when both main portal velocity and distal shunt velocity are abnormal, ultrasound has 100% specificity for detection of TIPS malfunction. Furthermore, it is important to follow these numbers over time because initially after TIPS, the velocities may all be normal, but changes in velocities and flow directions can indicate that a stenosis has developed. In fact, Dodd et al.  considered temporal changes in velocity more sensitive than static low-velocity parameters. Other investigators [13, 15–18] have supported that ultrasound is very sensitive for detecting shunt malfunction.
Unfortunately, not all investigators have found ultrasound to be especially useful. One prospective double-blinded study reported that ultrasound predicted shunt patency in 20 of 31 shunts that proved to be occluded or stenotic . However, their main criterion for calling a stenosis was a peak shunt velocity of < 60 cm/s, which is somewhat simplistic. Using more extensive velocity criteria, another study  still found concordance between ultrasound and venography in only 53% of cases, and in their experience ultrasound rarely predicted a stenosis that was not already suspected on clinical grounds. Given the variability in the reported sensitivity and specificity of ultrasound, it is important to evaluate the results at your own institution to see how your ultrasound readings correlate with venography and pressure measurements.
The use of echo enhancers has been proposed as a way of improving the accuracy of ultrasound. In a small study of 31 TIPS, the use of echo enhancers was found to increase the specificity of ultrasound from 89% to 100% . However, this technique has not become common practice.