Transjugular intrahepatic portosystemic shunt in portal hypertension: How to go further while staying on track?

Transjugular intrahepatic portosystemic shunt (TIPS) has now been available for over three decades, since the fall of the Berlin Wall.1 TIPS was first used for refractory bleeding from esophageal varices and then for the prevention of recurrent bleeding and recurrent tense ascites.2,3 It is quite interesting to note how the first pilot studies had not only shown the effect on portal hypertension (PH) complications but also confirmed that the shunt snags quickly (already described with surgical shunts). Subsequent randomized controlled trials and meta‐analyses showed that TIPS is the most effective method to treat complications related to PH.4,5 However, the risk of hepatic encephalopathy (HE) and the lack of a clear survival benefit have limited its expansion and popularity. However, the reason why this criterion is always required for TIPS is a matter of debate. Technical advances in the use of covered prostheses and improvement in the selection of patients have now made TIPS an undisputed procedure in the therapeutic armamentarium of patients with decompensated cirrhosis, and all the most recent guidelines support its use in recurrent ascites and as a preemptive therapy in acute variceal bleeding because of a significant survival benefit.6 Thus, the current perspectives concern, on the one hand, the increase in knowledge in the prevention and management of the risks related to the shunt, mainly HE and cardiac decompensation, and on the other hand, the extension of the indications to more patients (sicker, older) or to other situations (TIPS to prepare for surgery,7 TIPS in noncirrhotic PH8). These two objectives may seem contradictory. The invasiveness of TIPS has prompted the search for prognostic indicators to limit their use to patients with a low risk of shunting complications. However, in the present issue, the study by Vizzutti et al. proposes a predictive model of mortality after TIPS in patients with cirrhosis who are over 70 years of age, when these patients are at high risk of HE and cardiac decompensation.9 Another relevant contribution in this issue by Queck et al. reminds us of a key feature: the hemodynamic objective "portal hepatic systemic pressure gradient" (PPG) remains the main target for evaluating the result of the shunt at a time when everything is done to limit the risks of overshunting (smallest possible diameter, per‐procedure embolization).10 The aim of the first paper was to develop and validate a model for the prediction of mortality in older adults with cirrhosis treated with elective TIPS. Ninety‐nine older patients were included in the derivation cohort, and 76 were in the validation cohort. Exclusion criteria were Child‐Pugh score >11, Model for End‐Stage Liver Disease (MELD) score >18, bilirubin >5 mg/dl, creatinine >3 mg/dl, heart failure, porto‐pulmonary hypertension, recurrent HE, and uncontrolled sepsis. The primary outcome was liver‐related death. The probability of liver‐related death was higher in older patients (41% at 3 years vs. 21% in younger patients). Serum creatinine and serum sodium were the two only independent predictors associated with liver‐related mortality in older patients, and a prediction score was built on them (https://promisepa.shinyapps.io/TIPS/). The limits of the study are well discussed by the authors. Several points deserve to be highlighted. The overall Harrell's c‐indexes were quite low: 0.61 and 0.57 in the derivation and validation cohorts, similar outcomes to the most recent published models, which call for improvement. One way to achieve this would be to test new approaches (biomarkers of inflammation or circulatory dysfunction, radiological and/or cardiac parameters) combined with usual criteria and the use of new intelligent technologies. The model proposed by the authors only applies to already highly selected patients, which means that the parameters related to liver function are bound to be normal, mainly serum bilirubin and international normalized ratio, and that the present model cannot be used alone at the first step to select good candidates for TIPS. The frequent variations over time of serum sodium and creatinine values also make it difficult to choose the best window for the calculation of the score in a model with these two variables only. Another concern is the lack of a control group of the same age, which compromises the comparison of mortality but also of other important outcomes. One unexpected result is that there was no difference in the occurrence of HE after TIPS between older and younger adults, although age is one of the most frequent parameters associated with the risk of HE. We must emphasize that the assessment of HE remains subjective, and it is difficult to address it specifically in a nondedicated study. Most importantly, it is essential to collect data and make comparisons with the same group of age on outcomes such as HE, hospital readmissions, frailty, autonomy, and quality of life. For equal survival, these criteria are important to consider and may take precedence in older patients. The authors make a very helpful first step in better understanding outcomes for older patients; however, we still need more robust tools to risk‐stratify patients for an individualized medicine. For that, prognostic models must be derived from well‐designed prospective multicenter studies aiming to minimize the risk of many biases, including not only survival but also other outcomes of interest and considering the changes in the epidemiology of cirrhosis with the impact of comorbidities and the daily battle for removing factors (metabolic syndrome, alcohol use). The second study by Queck et al. published in this issue reminds us that targeted reduction of the PPG is the cornerstone of effective treatment of ascites in patients with cirrhosis. This single‐center retrospective study included 341 patients from 1994 to 2015 who underwent an elective TIPS procedure for the treatment of severe ascites. The aim of the study was to assess the effect of PPG reduction on patient outcome 6 weeks after TIPS implantation. Patients had a median age of 60 years, and the etiology of cirrhosis was chronic alcohol consumption in 71% cases. The mean Child‐Pugh and MELD scores were 9 and 12, respectively. It should be noted that both bare and covered stents were used. The most important finding was that patients without ascites at 6 weeks after TIPS had significantly greater PPG reduction compared with patients with persistence of severe ascites (65% vs. 55% of pre‐TIPS PPG; p = 0.001). A PPG reduction ≥60% was found to be the best threshold for controlling ascites. This result is reflected in an increase in survival. Lower PPG immediately after the procedure was also associated with resolution of severe ascites at 6 weeks. However, some questions remain to be answered. Is a 60% reduction in PPG the same in a patient with 14 mmHg before TIPS as in a patient with 30 mmHg? In this study, only 2% of the patients achieving a 60% PPG reduction still had a PPG above 10 mmHg. We assume that the absolute value of the final PPG rather than the percentage of PPG reduction should remain the main target to improve ascites and overall patient outcome. Does the persistence of clinically significant PH in patients with mild ascites (29% of the cohort) have an impact on prognosis? The median survival difference is more than 1 year between mild ascites and no ascites! How were patients with persistence of ascites at 6 weeks treated (diuretics, shunt dilatation…), and what was the impact on the outcome? Why did patients with an initial PPG below 10 mmHg after TIPS still have ascites at 6 weeks? In the absence of a PPG measurement at 6 weeks, we cannot know if this is the result of an early shunt dysfunction or if it is related to other causes. An unexpected observation is that HE improved at 6 weeks after TIPS. We should be cautious with this finding, and a close long‐term prospective follow‐up may temper this result. However, at a time when the focus is on the use of small‐diameter stents, in order to limit adverse events such as HE and heart failure, this study underlines the idea that the main objective is to decrease the PPG. This is the main parameter that has been shown to reduce PH‐related complications and increase survival. This is supported by all randomized studies assessing TIPS or drugs with an effect on PPG. A critical point is that it is challenging to target an "ideal PPG" in an individual case. One value is not necessarily suitable for all patients. The individual goal of abolishing PH‐related complications while limiting the risk of the shunt remains challenging. Furthermore, the difference between 7 and 8 mmHg, observed by the author regarding the resolution or not of ascites, is not conceivable for routine practice. It is illusory to aim to be accurate to the nearest millimeter of mercury. One should aim for a target area rather than a target value. Pending further studies, stent dilatation during the TIPS procedure still needs first to be adjusted according to the PPG results, which means that progressive dilatation should be performed until the hemodynamic target is reached (7–10 mmHg). In conclusion, if TIPS is now unavoidable in a small proportion of patients with cirrhosis, the present studies add a brick to the wall of knowledge, but some aspects still need to be clarified before other walls can be knocked down. This requires both large well‐designed prospective cohorts and randomized controlled trials.