Endoscopic ultrasound-guided portospenic split for refractory hepatic encephalopathy. Hepatic encephalopathy complicates about a third of patients with cirrhosis of liver during their lifetime and is associated with a high mortality, recurrent hospitalizations and poor quality of life. The mainstay of treatment are non-absorbable disaccharides like lactulose and antibiotics like rifaximin. However, a subset of patients continue to have persistent or recurrent symptoms. In these situations, therapeutic options are very limited. The patient may be listed for liver transplant, however, that may not always be possible because of logistical or financial reasons or because of delayed prioritization if the patient has a lower mild. Often these patients have large portosystemic shunts and may benefit from shunt occlusion. An alternative can be adjuvant therapies like branched amino acids, FMTs, however, the efficacy is limited. Shunt occlusion actually becomes the middle path for most of these patients. This is because gut-derived toxins like ammonia and lipopolysaccharides actually get shunted from the portal to the systemic circulation via the shunt, bypassing the liver and cause hepatic encephalopathy. These shunts can be blocked by interventional radiologists by a procedure called balloon retrograde transvenous obliteration where a balloon is inflated at the systemic end of the shunt and sclerosant is injected to obliterate the shunt. Alternatively, even endoscopic ultrasound can be used to obliterate these shunts by injecting coils and glue into the feeder to the shunt as has been demonstrated by us earlier this year. However, occlusion of shunts leads to decompensation of liver if the reserves are poor, that is, if the patient has jaundice, ascites or has a high mild score. So these patients can be very difficult patients to treat and such was the case that we encountered a 45-year-old male with ethanol-related cirrhosis who was referred to us because of recurrent hepatic encephalopathy for over a year. The patient had multiple breakthrough episodes of overt hepatic encephalopathy despite being on optimal doses of rifaximin, lactulose and branched-enamino acids. He had persistent asterisks at presentation, severe slurring of speech, making the speech unintelligible on many occasions even to family members. He also had hepatic parkinsonism and profuse facial dyskinesias which were extensively evaluated by neurologists and no other etiology was found. So this was also attributed to being a rare presentation of extrapyramidal features of hepatic encephalopathy. At presentation, the patient had thrombocytopenia, elevated bilirubin and elevated INR which brought his med score to 23. The patient was advised to go for a liver transplant, however, that was not an option because of financial constraints. So we had our backs against the ball. So this was the anatomy of the patient's portal circulation. There was a large shunt connecting the short gastric vein to the left renal vein. Occluding this shunt could have led to a reduction of photosystemic shunting and all of the blood from the mesentery as well as the spleenic circulation would have gone to the liver. However, the patient had a high meld and high bilirubin. So we expected that this much of an increased flow into the portal vein towards the liver may lead to hepatic decompensation. So with this, we were out of options considering transplant was off the table and the patient was already on optimal medical management. We thought out of the box here and realized that if we occlude the splenic vein, this would lead to a portal splenic split. That is, all of the splenic outflow would now go into the left renal vein and not congest the liver while the mesenteric circulation will still keep going into the liver and not cross over to the systemic side. So this seemed like a win-win to us. However, this was highly experimental and the exact thing was told to the patient and after discussion of therapeutic options, the patient opted to go ahead with this therapy. So here we have the echocardioscope at the stomach cardia. We can see the splenoportal confluence with the splenic vein taking off from the portal vein over here and right behind the neck of the pancreas, we can see this velocity of approximately 20-25 centimeters per second. And here we can follow this splenic vein towards the splenic hilum and we can see this large bend over here and going into the shunt. And now we are coming back towards the portal confluence. So we chose this site for the injection of coils, we punctured this with a 19 gauge needle and injected three 20 mm embolization coils followed by undiluted sinoacrylate glue, 2 cc's. This led to formation of a coil glue conglomerate over here. However, we still saw a fair bit of flow going towards the splenic vein. It was lesser than before. It is now down to 15-20 centimeters per second, but still significant. So we again injected the coil glue conglomerate with another 2 ml of undiluted sinoacrylate glue. And this did the job pretty well. And here we can see a large cast that is now seen in the splenic vein. And if we now go back to the portal venous end, we can see that the velocity of flow towards the splenic vein has now reduced to less than 10 centimeters per second. At this stage, we decided not to inject any more glue because these low velocities usually tend to thrombose on themselves. The patient did well post-procedure, barring mild abdominal discomfort for one day. That improved. Developed no ascites or no worsening of liver parameters and was discharged home on day 7. CT scan done a month later showed the presence of coils here in the splenic vein as expected with this cast in the splenic vein. We can see that the superior mesentric vein is continuing into the portal vein without any extension of thrombus and the splenic hilum is continuing into the splenorenal shunt. So we were able to achieve exactly what we expected to achieve here in terms of technique. Here we can again see that the splenic vein is completely occluded with coils and we can see a glue cast over here in the coronal and sagittal planes. Now this is the clinical improvement of the patient. The patient initially had asterexis and facial dyskinesias as are seen in the video. The video has been shared with consent of the patient. After a month, we can see that the asterexis has markedly improved and we see barely any facial dyskinesias now. The patient reported a much improved quality of life. At 3 months, the patient reported no episodes of overt hepatic encephalopathy. The speech is now clear. There is no more slurring and the family members can completely understand what the patient says. The asterexis is markedly reduced and extrapyramidal manifestations have also reduced markedly. So to conclude, refractory hepatic encephalopathy is a difficult situation to treat with very limited therapeutic options. Transplant may be the best option but may not be feasible for all patients. Shunt occlusion can definitely be the way to go for patients who have preserved liver reserves and a demonstrable shunt. However, in those who have poor liver reserves, options may be further limited. These patients may then benefit from a US-guided potospenic split. This will be feasible only if the shunt is a short gastric to left renal vein shunt because of obvious anatomical constraints. However, there is definitely a biological plausibility towards it. Currently, we have very limited data on this because previously only a couple of cases have been done by interventional radiology and that becomes very invasive. But this procedure is much simpler with endoscopic ultrasound guidance. However, the long-term results of this procedure are yet to be determined and it will take further prospective data to establish its place in the clinical scenario.

refractory hepatic encephalopathy

liver cirrhosis

splenic vein occlusion

endoscopic ultrasound

experimental procedure