Gastric varix (GV) is one of the lethal complication in patients with decompensated liver cirrhosis.^1-3 Because of increased variceal flow, hemostasis is difficult in GV. It can lead to rapid and massive bleeding.⁴ Endoscopic variceal obturation and balloon-occluded retrograde transvenous obliteration (BRTO) have been mainly used for prevention or treatment of GV bleeding.^5-10 BRTO is very effective in controlling GV bleeding. However, at least several hours of bed rest after the procedure is needed, and balloon rupture can lead to devastating outcomes.¹¹

It has been reported that BRTO improves liver function and hepatic encephalopathy (HE) of patients by blocking portosystemic shunt artificially and improving blood flow to the liver.^12,13 These procedures lead to improve the patient’s portal pressure than baseline, although with increased risk of portal hypertension related complications such as esophageal varix (EV) or ascites.⁸ However, few studies have examined the change in portal pressure in patients undergoing BRTO.¹⁴

Recently, plug-assisted retrograde transvenous obliteration (PARTO) has been widely used to compensate for the shortcomings of BRTO.^15,16 PARTO does not require an indwelling balloon, and entails a shorter procedure time than BRTO. Thus, it is advantageous for both the practitioner and the patient. Safety has been greatly improved using a plug rather than balloon to overcome the risk of balloon rupture on BRTO.^17-19 Although several safety and efficacy studies of PARTO are available, there are no serial follow-up studies investigating the changes in portal pressure after the procedure or its clinical impact on patients.

Thus, the purpose of this study was to compare the changes in portal pressure before and after PARTO and examine the incidence of related complications such as EV and ascites. This study also investigated the improvement in liver function and HE.

1. Patient population

A total of 73 patients underwent PARTO in a single tertiary hospital from March 2012 to February 2018. We retrospectively analyzed 54 patients, excluding 15 patients who were lost follow-up and four patients who died within a month after PARTO (two patients died from deterioration of liver function, the other two patients died from EV bleeding and infection, respectively). In the 26 of the 54 enrolled patients, hepatic venous pressure gradient (HVPG) measurement was performed before and after PARTO (Fig. 1). Among them, HVPG was measured incorrectly in one patient because of shunt between right and inferior hepatic vein. Therefore, the data of 25 patients were analyzed.

Patient characteristics are listed in Table 1. PARTO was performed for various indications (active bleeding, 11/54 [20.4%]; recent bleeding, 13/54 [24.1%]; primary prophylaxis for impending rupture 29/54 [53.7%]; and just for HE control without GV, 1/54 [1.9%]). One patients, who suffered from intractable and repeated disorientation for several months, had PARTO for HE control.

This study protocol conformed to the ethical guidelines of the World Medical Association Declaration of Helsinki and was approved by the Institutional Review Board of Soonchunhyang University Hospital (IRB number: 2019-03-024-001).

2. Techniques

One attending interventional radiologist performed the procedures. All patients were evaluated endoscopically for GV and indications for PARTO. Abdominal computed tomography (CT) was performed to confirm the presence of gastrorenal shunt or possibility of PARTO intervention. Patients were kept fasting and all medication potentially affecting portal pressure were discontinued 24 hours in advance.

The HVPG was measured and PARTO conducted after puncturing the right common femoral vein under ultrasound guidance and insertion of 7- or 8-F sheath (Terumo, Tokyo, Japan). We initially used the right common femoral vein in all patients. However, when insertion of a sheath into the gastrorenal shunt through the right common femoral vein was not feasible due to an acute angle between the left renal vein and adrenal vein, the right internal jugular vein was used. Right hepatic vein was selected using a 0.035-inch, 180-cm long-hydrophilic guide wire (Terumo, Tokyo, Japan). Wedged hepatic venous pressure was measured three times in the right hepatic vein with a 5-F, 7 mm balloon catheter (Clearstream technologies, Wexford, Ireland). The contrast material (Iomeron^® 300; Bracco, Konstanz, Germany) was used for inferior vena cava and wedged hepatic venograms. The balloon catheter was inflated until a similar wedged position was obtained. All procedures associated with catheter placement were carried out under fluoroscopic guidance. Adequate occlusion of the hepatic vein was confirmed by the absence of backflow following dye injection. Prior to recording of the tracings, the wedge hepatic venous pressure was stabilized for at least 45 to 60 seconds.²⁰

After the measurement of HVPG three times, the left adrenal vein was navigated through the left renal vein and the sheath was inserted into the gastrorenal shunt. The guidewire and catheter were subsequently removed, and the vascular plug was deployed. Among the various types of vascular plugs, we used the Amplatzer Vascular Plug II (AGA Medical Corp., Plymouth, MN, USA). The size of the vascular plug was selected according to the diameter of the narrowest gastrorenal shunt, as measured on a CT scan, and the sizes ranged from 8 to 20 mm in diameter or 20% larger than the targeted gastrorenal shunt so as to prevent migration (mean shunt size, 8.63±2.28 mm; mean plug size, 12.79±3.45 mm) (Table 2). To facilitate the entry into the gastrorenal shunt proximal to the vascular plug, the vascular plug was deployed at the most dilated gastrorenal shunt. The gastrorenal shunt proximal to the vascular plug was then negotiated using the guidewire and the 4-F angled-tip catheter between the vascular plug and the shunt wall. The vascular plug was pulled down to the narrowest gastrorenal shunt while maintaining the position of the catheter to block the shunt flow and to prevent vascular plug migration. When the varix was not properly identified by the feeding veins, the inflow veins were blocked with a tornado coil (Cook Inc, Bloomington, IL, USA) and a vortex coil (Target Therapeutics, Fremont, CA, USA), measuring 3, 4, and 5 mm using a 2.2-F microcatheter (Fig. 2A). The mean total procedure time was 70.13±35.69 minutes. HVPG was measured again after mounting the plug (Fig. 2B). The catheter was subsequently removed, and the delivery cable attached to the end of the vascular plug was then detached.¹⁹

3. Follow-up evaluation and definitions

Patients were assessed indirectly for changes in hepatic function and bleeding via laboratory testing on the day of PARTO and one day after PARTO. In the absence of any complication after PARTO, abdominal CT and endoscopy were performed within 1 to 3 months to evaluate the success of the procedure and re-evaluate GV. Patients had been followed up at intervals of 2 to 3 months and re-evaluated using laboratory tests and clinical findings. We reviewed the patients’ hepatic function at baseline, and again at 1 and 6 months after PARTO based on Child-Pugh score and Model for End-Stage Liver Disease (MELD) scores. We also evaluated long-term changes in patients with a mean follow-up time of 28±19.8 months (range, 6 to 78 months). The degree of ascites was measured via ultrasonography or CT scan and graded according to International Ascites Club guidelines.

The technical success of PARTO was defined by the complete occlusion of the efferent shunt and complete filling of GV with a gelfoam slurry. Eradication of GV was defined by complete or marked shrinkage of GV after PARTO on follow-up endoscopy or CT scan. The complications of PARTO were defined according to Interventional Radiology Standards of Practice Committee. The HE grade was determined according to West Haven Criteria. Liver function was compared with based on Child-Pugh score and MELD score. Serum albumin, bilirubin, and prothrombin time were also compared, respectively. For analysis of the factor affecting liver function, the definition of liver function improvement was considered based on an improvement in the combined patient’s both Child-Pugh score and MELD scores. EV deterioration was defined by bleeding or an increase in EV size compared with the conditions before PARTO.

4. Statistical analysis

All normally distributed data were presented as mean±standard deviation. Otherwise, data were presented as medians with interquartile range. The Student t-test or Mann-Whitney U-test was used for continuous variables, and the chi-square test or Fisher exact test was used for categorical variables to determine the difference between groups. The Wilcoxon signed-rank test was used to compare changes before and after PARTO. Logistic regression analysis was used to determine the risk factors for EV deterioration or liver function improvement. All statistical analyses were performed using IBM SPSS software version 22 (IBM Corp., Armonk, NY, USA). A p-value less than 0.05 was considered statistically significant.

1. Efficacy and safety of PARTO and portal pressure change

A technical success rate of 100% and a gastric variceal eradication rate of 92.6% (50/54) were achieved. Two of the four patients who were refractory to PARTO experienced additional GV bleeding within 12 months after PARTO and underwent endoscopic variceal obturation for emergent active bleeding control. Three patients experienced complication after PARTO and all were minor problems (Table 2). In 25 patients in whom the portal pressure was measured, there was a significant increase in portal pressure before and after PARTO (changing from 12.52±3.83 to 14.68±5.03 mm Hg, p<0.001) (Fig. 3).

2. EV change after PARTO and the relationship with portal pressure

Of the total 54 patients, forty-nine patients had esophagogastroduodenoscopy performed to evaluate EV deterioration. EV deterioration was observed in 26 patients (53.1%). There was no significant difference in EV deterioration when categorized by patient’s basal liver function (11/23 [47.8%] in Child-Pugh A patients; 15/26 [57.7%] in Child-Pugh B/C patients) (Table 1). Endoscopic variceal ligation was performed in two patients with EV bleeding within 1 month after PARTO. Propranolol requirement was increased for EV control (10/54 patients after 1 month, 12/54 patients after 6 months).

In 25 patients in whom the portal pressure was measured, EV was evaluated in 19 patients except in six patients who did not have follow-up endoscopy within 6 months after PARTO. Among 19 patients, EV deterioration was observed in 10 patients (52.6%) with a mean portal pressure change from 14.5 (11–16.5) to 17 (14.25–21.25) mm Hg compared with patients without changes in EV (nine patients [47.4%], HVPG from 11 [8–11.5] to 12 [6.5–15] mm Hg). EV deterioration was observed as bleeding (2/19 patients, 10.5%) or an increase in EV size (8/19 patients, 42.1%) compared with the conditions before PARTO, during the follow-up endoscopy within 6 months after PARTO. Only portal pressure after PARTO was a significant risk factor in multivariate logistic regression analysis (odds ratio, 1.341; 95% confidence interval, 1.017 to 1.767; p=0.037) (Table 3). Of 10 patients with EV deterioration, eight patients had post-PARTO portal pressure greater than 15 mm Hg. The other two patients were 9, and 12 mm Hg, and the patient who checked 9 mm Hg of HVPG showed 50% of portal pressure change after PARTO than baseline.

3. Changes in hepatic function after PARTO and the relationship with portal pressure

Both Child-Pugh score and MELD scores after PARTO showed significant improvement at 1 and 6 months (Child-Pugh score: baseline, 6.94±1.76; 6 months later, 6.30±1.44; p<0.001 and MELD score: baseline, 11.46±4.35; 6 months later, 10.33±2.96; p=0.021). However, hepatic function was not improved significantly during the overall follow-up period (Child-Pugh score: 6.87±1.84, p=0.781 and MELD score: 11.09±3.41, p=0.457) (Table 4).

Child-Pugh B/C group had a higher proportion of patients with improved liver function than Child-Pugh A group at 6 months after PARTO than baseline (4/26 [16.0%] in Child-Pugh A patients and 13/28 [46.4%] in Child-Pugh B/C patients). Among markers reflecting hepatic function, a significant improvement was detected only in serum albumin level (mg/L) in Child-Pugh A patients (baseline, 3.83±0.41; 6 months later, 4.03±0.55; p=0.012). However, in Child-Pugh B/C patients, a significant improvement of serum albumin level (mg/L) (baseline, 2.99±0.57; 6 months later, 3.46±0.68; p<0.001) and a decrease of prothrombin time/international normalized ratio level (baseline, 1.359±0.249; 6 months later, 1.280±0.244; p=0.004) were observed (Fig. 4).

The relationship between liver function and portal pressure was analyzed. The Child-Pugh score and MELD scores at 6 months were better than baseline in 10 patients (45.5%) (Child-Pugh score: 8 [6–9.25] to 5.5 [5–8]; MELD score: 12.5 [9.25–16.25] to 10 [8–14]). In multivariate logistic regression analysis, the portal pressure or its difference before and after PARTO was not a significant factor affecting liver function (HVPG of patients with improved liver function: 11.5 [10.75–17.25] to 15 [10.75–16.5] mm Hg; HVPG of patients without improved liver function: 12 [9.5–15] to 14 [11–21.25] mm Hg) (Table 5).

4. Changes in other cirrhotic complications after PARTO

Only four out of 54 patients had overt HE characterized by lethargy, disorientation, and mental deterioration with flapping tremor and higher blood ammonia levels (range, 135 to 340 mg/dL) at the time of admission to the hospital (one grade 1, one grade 2, and two grade 3). In all four patients, HE improved to grade 0 at 1 month after PARTO. Such improvement continued until 6 months after PARTO. No hospitalization or additional medication was required during the observation period. Among 48 patients with grade 1 or 2 ascites except for six patients with grade 3 ascites, an increase in the amount of ascites was detected in eight patients (16.7%) within 1 month and none at 6 months after PARTO.

The technical success rate (100%) of PARTO was similar with a recent study reporting 98.6% (72/73).¹⁵ The rate of complications related to the procedure was only 5%. In addition, all complications were minor such as fever and minor oozing at the puncture site. No serious complications such as bacterial peritonitis, portal or renal vein thrombosis, pulmonary embolism, or pulmonary edema were detected. These results support the safety and efficacy of PARTO.

EV is expected to deteriorate due to an increase in portal pressure after PARTO. According to BRTO studies, the risk of increased EV size and bleeding after BRTO was 7.3% and 27% within 1 year, respectively.^8,21,22 Jogo et al.²³ suggested that a serum total bilirubin level of 1.6 mg/dL or higher and an HVPG of 13 mm Hg or higher are risk factors for EV deterioration. According to a multicenter study of 183 patients in Korea, 54 of 136 patients (39.7%) showed worse EV grade after PARTO.¹⁰ In our study of 54 patients, EV worsening was observed in 26 patients (53.1%) and 16 of them underwent endoscopic variceal ligation. Of the 19 patients who had portal pressure measured and underwent follow-up endoscopy, only portal pressure after PARTO was a significant factor affecting EV deterioration. Ten patients with EV deterioration developed portal pressure of 15 mm Hg or more after PARTO. Of the remaining two patients, one patient had increased portal pressure by more than 50% from baseline. In multivariate analysis, portal pressure after PARTO had a significant effect on EV deterioration. However, sudden changes in portal pressure are expected to affect EV deterioration, but further study is needed with sufficient sample size.

The hepatic function of the treated patients improved significantly during the 6 months of observation period. In particular, the serum albumin level and the prothrombin time improved after 6 months compared with the baseline. A few studies on BRTO reported a transient improvement in liver function due to increased portal blood flow. Kumamoto et al.²⁴ reported that a large splenorenal shunt can improve the liver function compared with a non-BRTO-treated group. However, after an average of 3 years, the liver function became similar to the baseline. In our study, the liver function significantly improved during 6 months compared with that at baseline. However, it did not continue for long-term follow-up period during the overall 28±19.8 months (range, 6 to 78 months). Uehara et al.¹⁴ analyzed the changes in portal pressure following shunt occlusion and reported improvement in liver function during a 6-month observation period when a 20% increase over the baseline portal pressure was detected. And they reported patients with Child-Pugh class B or C showed more improvement in the albumin level, prothrombin time, and total bilirubin level compared to Child-Pugh class A. In the present study, the portal pressure change was not a significant factor in the improvement of liver function. In the patients with Child-Pugh class B or C, more patients improved liver function after PARTO. Among the parameter affecting liver function, serum albumin level and prothrombin time improved in the patients with Child-Pugh class B or C during 6 months after PARTO. However, only serum albumin level improved in the patients with Child-Pugh class A. Our results were similar to the previous study.

A major advantage of blocking the portosystemic shunt is the improvement in HE.^9,21,24-26 Of four patients, only one received PARTO just for HE control without GV. The improvement of HE was reported in the previous BRTO study as mentioned in the introduction,^12,13 and was based on the same results in the recent PARTO study in Korea.^15,19,27 All the four patients diagnosed with HE showed clinical improvement after PARTO in the present study. Overall, the serum ammonia level of the enrolled patients returned to normal range at 1 week after PARTO (from 127.4±58.0 to 28.1±9.8 µmol/L, p=0.002), similar to the results of other studies.¹⁹ In a study of ascites, the increase in ascites ranged from 0% to 44% of patients after BRTO.²⁸ In the present study, there was no clinically significant aggravation in ascites, especially at 6 months after PARTO.

This study has several limitations. First, this was a retrospective study. Thus, the interpretation of the results was limited in the absence of a control group. However, the results of this study support the safety and efficacy of PARTO, which has been performed widely, with excellent results for HE. The liver function also improved in these patients and was sustained over a mean observational period of 28±19.8 months (range, 6 to 78 months). However, a long-term follow-up is needed with larger sample size.

In conclusion, the portal pressure significantly increased after PARTO. During 6 months of observation period, hepatic function and HE improved, and EV deteriorated. Post-procedural portal pressure was the only significant risk factor for EV deterioration. Thus, a serial follow-up of portal pressure can facilitate the prediction of EV bleeding. However, further long-term follow-up and prospective studies are needed to corroborate these results.

This work was supported by the Soonchunhyang University Research Fund.

No potential conflict of interest relevant to this article was reported.

Guarantor of the article: Y.S.K. Drafting of the manuscript, acquisition and interpretation of data, statistical analysis: J.W.P., Y.S.K. Conduct of procedure and computed tomography reading: J.M.L., J.J.S. Study concept and design: S.G.K. Manuscript preparation and editing: J.J.Y. Critical revision of manuscript: S.W.J., J.Y.J., S.H.L., H.S.K., Y.D.K., G.J.C., B.G.J. All authors approved the final version of the manuscript.