Background
Gastric varices (GVs) are a complication of portal hypertension and are observed in approximately 20% of patients with liver cirrhosis [
1,
2]. They are less prevalent than esophageal varices (EVs) and are present in 5–33% of patients with portal hypertension, with a reported incidence of bleeding of approximately 25% in 2 years, and a higher rate observed in patients with fundal varices [
1,
3]. Although GVs bleeding is less frequent than EVs bleeding in patients with liver cirrhosis, bleeding from GVs is much more severe, requires more transfusion, and leads to a higher mortality rate [
4,
5]. The treatment for GVs includes endoscopy, surgery, interventional radiology using a catheter, and conservative therapy [
6‐
8]. The treatment of GVs bleeding also requires the involvement of a multidisciplinary team that includes hepatologists, endoscopists, diagnostic radiologists, and interventional radiologists [
9].
Balloon-occluded retrograde transvenous obliteration (BRTO) is an interventional radiological treatment for GVs that was originally introduced in Japan [
10‐
12]. BRTO is reported to be an effective strategy against GVs with a gastro-renal shunt, and reportedly results in a high GVs obliteration rate and low recurrence rate [
13‐
18]. However, exacerbation of EVs is common (seen in 13–56% of cases after 1 year) after GVs eradication by BRTO, and several predictive factors are reported to be associated with the exacerbation of EVs [
14,
16‐
20]. Hepatic venous pressure gradient (HVPG), which is calculated by measuring the difference between wedge hepatic venous pressure and free hepatic venous pressure, has been recognized as a reproducible and reliable method for assessing portal pressure and as a predictive factor for the development of EVs in patients with cirrhosis [
3,
21‐
23]. However, few reports have described the clinical significance of HVPG in the prognosis of patients who have undergone a BRTO procedure and show EVs exacerbation [
20,
23].
In the present study, the relationship between long-term prognosis, exacerbation of EVs, and portal hypertension estimated by HVPG was analyzed to identify predictive factors related to the prognosis of patients undergoing BRTO for GVs, and the incidence of EVs exacerbation in these patients.
Discussion
GVs are complications of portal hypertension and are less prevalent than EVs [
1,
3]. However, bleeding from GVs is much more severe than that from EVs and leads to a higher mortality rate [
4,
5]. Thus, treatment of GVs caused by portal hypertension is important when treating patients with liver cirrhosis. BRTO is an interventional radiological treatment for GVs with gastro-renal shunts that has been reported to provide a favorable GVs obliteration success rate (92.1–97.4%) [
15‐
18]. On the other hand, exacerbation of EVs has been reported to occur after BRTO due to portal hypertension, with a cumulative exacerbation rate of 8.6–27% at 1 year, 20–58% at 3 years, and 18–66% at 5 years after BRTO [
15‐
18]. Obstruction of a large gastro-renal shunt and the change in hemodynamics caused by BRTO appear to lead to the aggravation of EVs, although the mechanism of the exacerbation remains unclear [
16]. Waguri et al. reported that concomitant partial splenic embolization (PSE) with BRTO might contribute to prevention of the exacerbation of EVs after BRTO [
27]. In the present study, however, the phenomenon could not be analyzed because none of the cases underwent concomitant PSE with BRTO. In terms of portal vein thrombosis, three out of the 18 patients with HCC had portal vein tumor thrombosis, among whom only one patient had exacerbation of EVs after BRTO. Additionally, there was no statistically significant difference in the cumulative exacerbation of EVs with or without portal vein tumor thrombus. According to several reports, including our previous report, co-existence of EVs was a predictive factor for the exacerbation of EVs after BRTO [
15‐
17]. In the present study, the co-existence of EVs was not related to the exacerbation of EVs after BRTO, despite the fact that the co-existence rates were similar to those observed in previous reports (63.9 vs. 45.6–68.0%) [
15,
17].
HVPG has been widely used to assess portal pressure and the predictive value of EVs development in cirrhotic patients [
3,
21,
22]. BRTO has been reported to significantly elevate HVPG [
23,
28], as was also observed in the present study. Jogo et al. reported that an HVPG of ≥13 mmHg before BRTO is an independent risk factor for EVs aggravation after BRTO [
20]. Additionally, there are some reports that elevation of HVPG after BRTO is related to the aggravation of EVs after BRTO [
23,
29]. In the present study, post-HVPG correlated more significantly with EVs exacerbation after BRTO than pre-HVPG or the elevation of HVPG. Since obstruction of large gastro-renal shunt vessels is not inevitable during the BRTO procedure, post-HVPG might be a much more reliable parameter for monitoring of EVs exacerbation over an extended time period and might closely reflect portal hypertension and hemodynamics after BRTO.It has been reported that HVPG <12 mmHg and a reduction in HVPG by ≥20% of the baseline value reduces the risk of bleeding and death in cirrhotic patients with portal hypertension [
30]. In the present study, as well, patients with post-HVPG ≥13 mmHg tended to have a poor prognosis. This suggests that a high portal pressure after BRTO might increase the risk of variceal or non-variceal complications in patients with GVs.
Left gastric vein (LGV) diameter and LSM have also been reported to be important risk factors for worsening of EVs after BRTO [
31,
32]. The present study failed to show the correlation between LGV diameter and exacerbation of EVs, although the LSM value showed a correlation with EVs exacerbation after BRTO, with patients with high LSM values being more likely to develop exacerbation of EVs. LSM is useful for the non-invasive assessment of not only liver fibrosis, but also portal hypertension [
33]. Both liver stiffness and splenic stiffness strongly correlate with HVPG in cirrhotic patients with large esophageal varices [
34], and LSM positively correlates with the presence of EVs in patients with HCV-related cirrhosis [
35]. Ogasawara et al. reported that LSM at 24 weeks after anti-HCV therapy is a predictor of EVs exacerbation after achieving SVR in patients with HCV-related cirrhosis [
36]. In the future, studies involving a larger number of patients and more detailed analysis of the interrelation between HVPG, LGV diameter and liver and spleen stiffness are required to clarify the predictive factors and prevent the exacerbation of EVs after BRTO.
In the present study, patients with high LSM showed a poor prognosis after BRTO. Ishikawa et al. reported that LSM < 21.6 kPa predicted improvements in Model for End-stage Liver Disease (MELD) sodium scores and in survival rates after portosystemic shunt occlusion by BRTO [
37]. The present study showed a similar cut-off value of LSM; patients with LSM < 21 kPa had a more favorable prognosis compared to those with LSM ≥ 21 kPa. As previously described, LSM seems to reflect both liver fibrosis and portal hypertension. Therefore, LSM might perceptively predict the exacerbation of EVs and prognosis after BRTO, and patients with high LSM levels seem to need careful consideration for possible subsequent exacerbation of EVs and poor prognosis after BRTO.
The present study has certain limitations. First, is the small sample size and the short observation periods. Although the study included only 36 patients, our results might provide basic clinical data for further large-scale cohort studies or validation studies aimed at identifying predictive factors of unfavorable outcomes in patients undergoing BRTO. Second, there might have been a selection bias. The present study included two patients with Child–Pugh class C cirrhosis. Whether or not BRTO should be introduced in patients with poor hepatic functional reserve, such as those with Child–Pugh class C cirrhosis, depends on the patient’s preference and clinician’s discretion. In addition, cases that underwent treatment for GVs bleeding were included, which might have affected the study results. Thus, further analysis is needed to clarify factors associated with the prognosis after BRTO using a larger number of patients, including those with poor hepatic functional reserve or GVs bleeding.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.