My colleagues and I were pleased to see the prospective studies by Sanyal [1] and Cello [2] and their colleagues comparing transjugular intrahepatic portosystemic shunts with endoscopic sclerotherapy for the control of variceal bleeding. We were disturbed, however, to see such a high incidence of recurrent variceal hemorrhage in the TIPS populations. In Sanyal and colleagues' group, 9 of the 39 patients who had a TIPS placed returned with variceal hemorrhage. Six of the 9 patients died of their variceal bleeding.

We believe that a 23% rate of recurrent variceal bleeding within 1000 days after TIPS is unacceptably high and is probably related to the manner in which the patency of the intrahepatic shunt was assessed. Although current reports have suggested that Doppler ultrasonography is sensitive in detecting hemodynamically significant intrahepatic shunt stenosis and occlusions, the accuracy of this method has never been proven in a prospective, double-blinded study using preset ultrasonographic criteria.

My colleagues and I, like Sanyal and coworkers, had been using real-time gray-scale, color flow, and Doppler imaging to assess the patency of intrahepatic shunts. We stopped using this method, however, when three near-fatal variceal bleeding episodes occurred in our follow-up TIPS population. All three patients redeveloped portal hypertension from stenosis of the intrahepatic shunt, and all three patients had ultrasonography on admission that failed to detect any evidence of shunt malfunction. One patient developed massive variceal bleeding 1 week after ultrasonography of his shunt had yielded a negative result. We then decided to conduct our own prospective double-blind study comparing real-time gray-scale, color flow, and Doppler ultrasonography with direct venography and portosystemic pressure measurements. We found that at our institution, the sensitivity of ultrasonography for detecting hemodynamically significant shunt stenosis or occlusion was less than 50% compared with venographic evaluations.

Since we have returned to shunt venography and direct portosystemic pressure measurements for reassessment of shunt patency, no major recurrent variceal hemorrhage has occurred in our TIPS population. We reassess the TIPS at 1, 3, and 6 months and then every 6 months after the primary TIPS procedure and revise the shunts as needed. Venography and shunt revisions are performed on an outpatient basis. Venography is more invasive than ultrasonography, but in more than 200 reevaluations performed with venography, no complications related to the shunt reassessment or revision have occurred.

In summary, we have not found ultrasonography to be sensitive in the detection of intrahepatic shunt stenosis. This may explain the unacceptably high incidence of fatal variceal bleeding seen in the cited studies. We believe that the TIPS procedure as it is currently performed requires close follow-up. Although venography costs more, the price of missing an underlying shunt stenosis more than offsets any savings that may be achieved with ultrasonography.