Identifying the sweet spot in portal pressure reduction with TIPS

In 1990, Roberto Groszmann published a seminal paper identifying a threshold value of 12 mm Hg as measured by wedge hepatic pressure measurement, below which the risk of variceal bleeding was virtually eliminated.1 For over 30 years, this value became a target for pharmacological and radiologic intervention, and countless studies have continued to confirm that bleeding is extremely rare at portal pressure values <12 mm Hg. The difficulty arises if this value is not met. During TIPS procedures, the response is then either further dilatation of the stent or even placement of a parallel stent,2 both maneuvers associated with the subsequent increased risk of HE. Percentage portal pressure reductions may provide a more individualized patient approach. A 50% reduction in portal pressure at the time of the TIPS procedure significantly reduces portal hypertension–related complications and is the recommendation included in the Baveno 7 consensus document.3 However, lower portal pressure reductions, ranging between 10% and 20%, have also been shown to reduce bleeding.4–6 Underdilatation of a TIPS stent risks increasing the risk of bleeding/ascites recurrence but may also reduce the incidence of post-TIPS complications such as encephalopathy.7 In this month’s Hepatology, there are 2 studies aiming to define whether there is a “sweet spot” of portal pressure reduction by elective TIPS that decreases portal hypertensive bleeding and ascites without inducing negative complications such as HE. The conclusions are contrasting. Roccarina and colleagues present a large Italian multicenter retrospective analysis of the impact of elective TIPS end-procedural target pressure adherence. The authors identified 203 patients who underwent TIPS for secondary prophylaxis against variceal bleeding and 212 patients who had TIPS for diuretic-resistant ascites. The numbers undergoing TIPS for secondary bleeding prophylaxis are perhaps surprising given the conflicting evidence for this approach,8 but this does not detract from the paper itself. The decision process regarding which patients were underdilated is not established. Fewer patients in the ascites arm achieved a hemodynamic response suggesting perhaps that the clinicians were already modifying their practice in this subset of patients, but the numbers underdilated were equal across both groups: 51% in the bleeding arm and 50% in the ascites arm had their tract dilated with a balloon diameter ≤7 mm. Essentially, failing to achieve the 2 preidentified portal pressure targets (>12 mm Hg, or >50% reduction) did not result in an increased number of bleeding events. While ascites were slower to resolve in those patients who did not achieve the hemodynamic targets, after 1 month, there was no difference. Kabelitz and colleagues present a second study from 3 European centers in Germany and Austria. Using a machine learning tool, they identify the optimum portal pressure gradient (PPG) reduction to achieve ascites resolution without increasing overt encephalopathy incidence. From a derivation cohort of 438 patients, the authors conclude that an optimal portal pressure reduction of 60%–80% was associated with reduced recurrent ascites without significantly increasing HE, tested on a validation cohort of 291. While both studies are retrospective, one suggests failing to achieving pre-set targets has little impact on clinical outcome and the other says it most certainly does. If one focuses on the patients with ascites in both studies, are there differences to explain this? Patients in the Italian cohort were older and more likely to have metabolic-associated steatotic liver disease (17% vs. 7%), both known risk factors for HE after TIPS. Pre-TIPS PPG was higher (22 vs. 18 mm Hg) as was post-TIPS PPG (11 vs. 7 mm Hg). Both studies had a high incidence of grade 2+ HE before TIPS (23%), with post-TIPS HE higher in the Italian cohort (between 40% and 55%, the higher figure in patients achieving a higher hemodynamic response). Are these cumulative differences enough to explain the different results— potentially, yes? A younger patient with more alcohol-associated disease may better tolerate a larger shunt. In addition, issues such as cessation of alcohol can have a significant impact on both portal pressure but also HE. For the hepatologist referring to radiology, the issue of underdilating stents is complex and involves issues of both stent material and technology. Underdilating a fixed diameter nitinol stent usually results in the end diameter being that of the stent's intended diameter, that is, an 8 mm stent dilated to 6 mm ends up being 8 mm. This is due to the radial forces inherent in this material and is a recognized phenomenon outside of the liver but has also been shown in TIPS procedures.9 The portal vein can be quite fibrous and may resist the stent radial force; hence, the expansion response in this situation is not predictable. Along similar lines, stent technology has also developed. Older nitinol stents dilated to 7 mm may eventually reach 10 mm whereas new controlled expansion stents dilated to 7 mm will reach 8 mm. Therefore, the same intervention on paper may have had different physiological impacts in the 12–23-year analysis period in both studies. An opportunity perhaps missed in the Italian study was that the authors did not examine the impact of other portal pressure targets (eg, 10% or 20%). Furthermore, in both studies, portal pressure was taken at the time of the TIPS procedure. It is recognized that deep sedation may influence portal pressure measurement,10 but there is also evidence from 2 studies that the pressure measurement taken at the time of the TIPS does not reflect the portal pressure measured subsequently. A Spanish study reported on a retrospective cohort of 155 consecutive patients, with 3 time points of pressure measurement-immediate, early, and late.11 The study group was heterogeneous: 36 patients were emergency, it included patients with Budd Chiari and PVT and over 50% of the patients were on vasoactive drugs at the time of the TIPS. Patients had immediate, early (24 h), and late (1 mo) portal pressure measurements. Essentially, the immediate measurement did not correlate with the early measurement. If sedated, the immediate value was lower; if awake, then it was higher. The early measurement correlated with the 1-month measurement. A consistent PPG of <12 mm Hg appeared to protect against portal hypertensive events, although the incidence of encephalopathy is not provided. A Chinese study12 explored prospectively the optimal timing of portal pressure measurement in 466 patients with variceal bleeding. The procedure was performed under local anesthetic (this, along with the number of patients, is noteworthy). Eighty-eight percent were elective/secondary prophylaxis, and in these, beta-blockers were stopped 24 hours before the procedure. Variceal or shunt embolization was performed BEFORE tract dilation and stent deployment. As per the Spanish study, the immediate value did not correlate with the 24-hour value. The 24-hour value correlated with portal hypertensive complications, and the authors suggest that a target PPG of between 11 and 14 mm Hg was associated with reduced HE, but without compromise of efficacy. A high incidence of PVT (37%) does confuse the issue somewhat as the portal vein will remodel over time and so later pressure drops are more of a function of clot resolution as opposed to stent function. How does one take this information forward?”. While the results from Kabelitz and colleagues favor quite a significant reduction in pressure (60-80%) in the discussion, they support a more cautious, step-wise approach when TIPS is placed in patients with ascites. In many respects, the Italian paper supports that approach, that is, it does appear that at the time of TIPS insertion, less is almost certainly more. A minor reduction in portal pressure, identified at the time of the procedure, is probably going to be enough. In the context of variceal bleeding, if there is no variceal filling on a splenic run at the end of the procedure, then that is likely to be sufficient. If varices persist, and there is good flow in the TIPS, then embolization rather than further dilatation may be a better way to go. In practice, when TIPS is performed for ascites, we routinely dilate to 6 mm and are happy with just a 2–4 mm Hg pressure reduction as long as there is good flow. If ascites do not resolve, one can always come back and dilate further. If encephalopathy occurs, this seems much harder to reverse as the patients quickly lose ground physically. In this respect, we are in agreement with Roccarina and colleagues—while post-TIPS HE has been suggested not to impact survival, anyone who has seen these patients in the clinic and observed the impact on relatives and caregivers will be under no illusions as to the impact this has. Should we be performing pressure studies 24 hours after the procedure? This would seem to be the gold standard, and while academically attractive, it will add substantially to the cost of the procedure and length of stay. In terms of portal pressure targets, a gradient of <12 mm Hg remains protective against complications of portal hypertension. How one “dials up” the optimal portal pressure reduction requires a form of controlled expansion type stent, which would allow one to dilate to 6, 8, or 10 mm. It is also up to us as hepatologists to select the right patients. There are a lot of patients having TIPS for secondary prevention of variceal bleeding despite an absence of good evidence to support this, and clearly, new prospective trials are required in this area. In addition, as Roccarina and colleagues point out, the bleeding patient is different from the patient with diuretic-resistant ascites. The latter is much further down the line of decompensation, and the post-TIPS oHE rates in these studies were high (35%–55%), questioning whether we are selecting the right patients. When one considers non-transplant candidates, TIPS for the management of ascites is performed for symptom control, then outcomes MUST include an assessment of the quality of life.