Supplementary Protocol

1. Patients

Out of 224 patients treated with EIS, 88 patients from whom we could obtain written informed concent, were enrolled in this study. Patients were confirmed absence of hepatocellular carcinoma. The inclusion criteria of EIS were as follows: EGV evaluated as F2 or F3 or red color sign on endoscopy, Child-Pugh classification of grade A or B, performance status of grade 0 or 1 and absence of refractory ascites.

2 Xe-CT imaging protocol

We used 25 % stable Xe gas in conjunction with an AZ-726 Xe gas inhalation

system (Anzai Medical, Tokyo, Japan). The wash-in and wash-out periods were both 4 min. The entire liver was CT-scanned at 1-min intervals at 4 levels, including the porta hepatis (9 scans in total, including the baseline scan). Using an AZ-7000W image processing system (Anzai Medical, Tokyo, Japan), the LV, portal venous tissue blood flow (PVTBF), and hepatic arterial tissue blood flow (HATBF) were calculated, and PVTBF and HATBF maps were created. Total hepatic tissue blood flow (THTBF) was calculated as the sum of PVTBF and HATBF, and THTBF maps were also created. The time course change rate for the arterial Xe concentration, which was needed to calculate PVTBF and HATBF, was derived using the time course of the Xe concentration in spleen tissue. An Aquilion CT scanner (Toshiba Medical Systems, Tokyo, Japan) was used, withexposure factors of 120 kV, 150 mA, and 13.8 mGy. All examinations were performed with the patients in the fasting state. Informed consent was obtained from each patient. Xe-CT was performed on admission and within a week after completion of EIS. All study protocols were reviewed and approved by the ethics committee at the St. Marianna University Hospital (approval no. 480).

3 EIS protocol

The concept of our endoscopic injection sclerotherapy technique is embolization of feeding vessels of esophageal varices within portosystemic collaterals by injecting sclerosant.Before endoscopic injection sclerotherapy, each patient was premedicated with an intramuscular injection of 0.5% atropine sulfate (TERUMO CORPORATION, Tokyo, Japan), 15 mg of pentazocine (PENTAGIN®, Daiichisankyo Pharmaceutical Co., Tokyo, Japan), and 25 mg of hydroxyzine hydrochloride (Atarax-P®, Pfizer Japan Inc. Tokyo, Japan). Lidocaine jelly or spray (Xylocaine® jelly, Xylocaine® pumpspray, Astrazeneca, Osaka, Japan) was applied to the pharyngeal area as a topical anesthetic. A balloon, referred to as the oral side balloon in this study, was attached to the tip of an endoscope (GIF-XQ 260, Olympus, Tokyo, Japan) and inflated as the contrast medium, iopamidol (Iopamiron®, Bayer Schering Parma, Osaka, Japan) was injected to prevent the sclerosant, 5% ethanolamine oleate (OLDAMIN®, ASKA Pharmaceutical. Co., Ltd., Tokyo, Japan), from flowing out of the varices into the systemic circulation. The sclerosant, 5% ethanolamine oleate, was injected using a combined intravariceal and paravariceal technique. After the start of injection of the sclerosant into the varices, the flow of the sclerosant was monitored using x-ray fluoroscopy. The injection of the sclerosant was stopped just as it filled the portosystemic collaterals. However, embolization of the feeder could not be achieved when variceal puncture or accidental retraction of the needle from the varices occurred during injection of the sclerosant. Puncture needles ranged in size from 23 to 25 gauge according to the size of the pore for the biopsy forceps of the endoscope. A maximum total volume of 20 mL of sclerosant was injected at any one endoscopy session for prevention of variceal bleeding. The sclerotherapy session were performed every one week. Subsequent sclerotherapy was undertaken at one-week intervals until the varices were eradicated. As necessary, additional treatment with polidocanol(Aethoxysclerol®, Kaigen CO., Ltd, Osaka, Japan) was applied at F0 on the basis of the endoscopic findings.

4 ICG R15

ICG R15 (%) were examined using ICG (Diagnogreen®, Daiichisankyo Pharmaceutical Co., Tokyo, Japan; 0.5 mg/kg body weight) on admission and within a week after completion of EIS.

5 Statistical analysis

All statistical analysis are performed by statistical software “GraphPad Prism” (GraphPad Software, California, USA).