Transesophageal Intrahepatic Portosystemic Shunts Compared with Sclerotherapy

IN RESPONSE:

We thank Granados and colleagues and Dr. Owens for their interest in our work. They raise several important issues. We agree with Dr. Owens that Doppler ultrasonography is not as sensitive as angiography for the detection of stent stenosis. In 1991, when our study begin, almost nothing was known about the natural history of portal hypertension after TIPS creation. As reported in our paper, we therefore did Doppler ultrasonography 1 and 7 days after TIPS and 1 month and every 3 months thereafter. Angiography was performed every 6 months or if clinical ultrasonographic evidence of recurrent portal hypertension was present. The reported instances of variceal bleeding occurred despite this fairly intense monitoring protocol.

We have also published a direct comparison of angiography and ultrasonography for the detection of stent stenosis [1]. With angiography used as the gold standard, Doppler ultrasonography was a specific (specificity, approximately 90%) but relatively insensitive (sensitivity, 70%) tool for detecting stent stenosis. We therefore agree that angiography is the ideal way to diagnose stent stenosis and the only way to treat it. Because most rebleeding occurs within the first 6 months after TIPS creation, I believe that Dr. Owens's approach of angiographic detection and treatment of stent stenosis at 3 months may be helpful in preventing such rebleeding. However, angiography performed at 1 month is likely to be a low-yield procedure for detecting stent stenosis because development of the pseudointima usually takes 6 to 8 weeks.

We also agree with Granados and colleagues that studies vary considerably with regard to severity of underlying disease, time from active bleeding, the use of concomitant β-blockade, and the schedule and technique of sclerotherapy. In our study, 7 of 12 deaths in the TIPS group occurred in patients with Child-Pugh class C cirrhosis compared with 5 of 7 deaths in the sclerotherapy group. As noted in our article, however, this finding was not significant when included in a proportional hazards model.

The timing of randomization is a critical determinant of risk for rebleeding and survival. We were primarily interested in the long-term effects on rebleeding and therefore randomly assigned patients after the greatest risk period for early rebleeding had passed. In that respect, our study varied significantly from the other published studies. Although salvage with TIPS may have contributed to the improved survival in the sclerotherapy group, we were unable to demonstrate this statistically by using interventions for rebleeding as a variable in a time-dependent proportional hazards model. However, we agree that, from a clinical point of view, TIPS certainly is a useful salvage procedure for patients in whom endoscopic treatment fails.

Finally, we completely agree with Granados and colleagues' comments about the value of TIPS in the setting of active bleeding that fails to respond to endoscopic therapy. We studied the utility of TIPS in this setting and published our findings in 1996 [2]. Briefly, patients with exsanguinating variceal hemorrhage who were considered to be at high risk for death after surgery (with such conditions as aspiration pneumonia, renal failure, sepsis, deep coma, and tense ascites) were treated by TIPS. Hemostasis was achieved in 29 of 30 cases, and the 30-day overall survival was 63%. We therefore concur that TIPS is an excellent salvage treatment for patients with active bleeding. However, it was not our specific aim to examine this topic in our recent Annals article.

• Copyright ©2004 by the American College of Physicians