This question recently came up on wards regarding a patient with early cirrhosis, otherwise compensated, who would abruptly, without identifiable precipitants, develop hepatic encephalopathy.
The issue was nicely addressed in a review published in Therapeutic Advances in Gastroenterology:
Spontaneous portosystemic shunts in liver cirrhosis: new approaches to an old problem
According to the review, this condition is common in patients with cirrhosis and correlates strongly with the development of hepatic encephalopathy. IR embolization of spontaneous shunts may be an effective treatment, depending on the anatomy, particularly in those patients whose underlying liver disease is relatively mild and whose episodes of encephalopathy are abrupt and without identifiable precipitants.
An algorithm for patient selection is presented. In the algorithm, if patients respond well to standard treatment no further investigation is necessary. Otherwise, consideration for IR embolization is recommended in those with a MELD of 11 or less. Imaging in search of shunts should take place in such patients. Should shunts be found, if anatomically feasible, IR embolization should be considered. Patients with portal hypertension related complications would be excluded. In patients with a MELD of 12 to14 decision making should be individualized. In patients with MELD of 15 or above transplantation is the preferred treatment.