Measurement of Inferior Vena Cava to Shunt Distance in Deciding Access Route for Balloon-Occluded Retrograde Transvenous Obliteration Procedure: A Pilot Study

INTRODUCTION

Balloon-occluded retrograde transvenous obliteration (BRTO) is performed to treat failed medical and endoscopic management of bleeding gastric varices and blocks shunts causing hepatic encephalopathy.^[1,2] Various transcatheter agents have been described to treat these shunts. These range from coils, plugs, sclerosants, and gelfoam. Occlusion balloons are placed in the shunt either through a femoral or jugular vein access, through which these agents are deployed. In case of sclerosants, once the foam is injected, it causes endothelial damage and thrombosis of the shunt. Femoral and jugular routes are used, with the former being preferred. Gastrorenal shunt is the most common shunt blocked during a conventional BRTO procedure, and in 90% of patients with gastric varices, it provides venous outflow. Gastrocaval shunt and transdiaphragmatic veins provide venous outflow in the remaining 10% of patients with gastric varices. Most operators use transfemoral approach and few use transjugular approach exclusively. Pre-procedural cross-sectional imaging helps to decide the approach,^[3] but it can be difficult to decide which route is most suitable for the BRTO procedure. Technically difficult (instability of guide wire in shunt and longer duration between access and shunt negotiation) shunt cannulations were thought to be due to an unfavorable angle between gastrorenal shunt and renal vein. However, we propose that inferior vena cava (IVC) to shunt distance also contributes in determining the preferred route for a BRTO.

MATERIALS AND METHODS

The study was approved by the Institutional Review Board. Informed consent was obtained for the procedure in all cases. We identified 16 patients who had undergone BRTO procedures with pre-planned contrast-enhanced computed tomography (CT) from June 2015 to March 2018 and retrospectively reviewed the clinical record, imaging, biochemical parameters, and clinical outcome. In four patients, the initial approach (two femoral and two jugular) failed due to unfavorable anatomy related to the distance between the shunt and IVC. Subsequent attempts using the alternative approach were successful in all four patients.

All measurements were done before BRTO procedure, and the procedures were performed by three interventional radiologists of 15, 10, and 7 years of experience. Coronal maximum intensity projection (MIP) images were created using portal or hepatic venous phase images; either on the IntelliSpace Portal DX Server, Philips Healthcare [CAD2], or RadiAnt DICOM Viewer, Medixant, Poland. Measurements were calculated either in true coronal or oblique coronal MIP images, but both IVC and shunt had to be visualized in same plane. IVC to shunt distance (ISD) was measured from the distal aspect of renal vein IVC junction to the proximal aspect of shunt to renal vein junction (Figure 1). We did not correlate these values with the venogram images. The authors felt that this had the potential to introduce errors due to variability in defining the distal aspect of renal vein-IVC junction because of renal vein contrast mixing with infrarenal IVC unopacified blood.

Close

Figure 1

Method to measure inferior vena cava to shunt distance.

Export to PPT

RESULTS

Sixteen cases of BRTO were undertaken between June 2015 and March 2018. Eleven patients had gastric variceal bleeds and five patients had recurrent hepatic encephalopathy. Most of them had gastrorenal shunts from the left gastric vein to the left renal vein. Two patients with hepatic encephalopathy had more than one shunt. There was male predominance, with a male-to-female ratio of 15:1. The mean age was 70 years (range 23–80 years). Table 1 summarizes the patient details, cause for bleeding, type of shunt, IVC to shunt distance, route of access, and outcome.

Twelve patients had isolated gastrorenal shunt. Two patients with refractory hepatic encephalopathy had both gastrorenal and splenorenal shunts with separate renal vein opening. One patient with refractory encephalopathy had isolated splenorenal shunt. One patient with gastric variceal bleed had both gastrorenal and splenorenal shunts with common drainage into the left renal vein. When measured on reformatted true and oblique coronal CT scan MIP images, the ISD was between 2 and 3.5 cm in 11 patients and >3.5 cm (3.7, 4.3 and 3.82 cm) in 3 patients. Two patients with more than one shunt had ISD <3.5 cm for gastrorenal shunts and >3.5 cm for splenorenal shunts.

The femoral approach was used in 11 patients, jugular approach in 3 patients, and both approaches for 2 patients. The jugular approach was successful in 3 patients with ISD >3.5 cm; however, femoral approach failed to cannulate the shunt successfully in 2 of these patients (Figures 2 and 3). In one of these patients with a longer ISD, the shunt was embolized by jugular approach first without attempting femoral approach.

Close

Figure 2

66-year-old male, known Hepatitis B virus cirrhosis, presented with recurrent HE (a) Contrast-enhanced computed tomography coronal maximum intensity projection image in a patient with recurrent hepatic encephalopathy, who had undergone unsuccessful coil occlusion of the shunt. The gastrorenal shunt is poorly seen because of the metal artifact from the coils. Large gastric varices are seen in the gastric fundus. (b) Left renal venogram showing the patent left renal vein and previously placed coils. Balloon-occluded retrograde transvenous obliteration (BRTO) failed due to the inability to catheterize the shunt from a femoral approach. (c) Fluoroscopic image (right anterior oblique projection) taken during deflation of the balloon following coil-assisted BRTO from the jugular approach shows coils and shunt occlusion.

Export to PPT

Close

Figure 3

57-year-old male, known NASH cirrhosis, presented with active gastric variceal bleeding. (a) Contrast-enhanced computed tomography coronal maximum intensity projection image shows a small gastrorenal shunt with ISD of 3.69 cm and mid caval narrowing above and below the hepatic segment of inferior vena cava (IVC). Angle between shunt and renal vein is obtuse. Vertical line drawn along the IVC margin measures 4.71 cm. (b) Fluoroscopic image shows sheath in renal vein and catheter in shunt. (c) Fluoroscopic image during successful balloon-occluded retrograde transvenous obliteration demonstrating staining of shunt. Easy negotiation of balloon occlusion catheter into shunt through transjugular access.

Export to PPT

In two patients with ISD 2.77cm (bleed) and 3.13 cm (bleed with two shunts), we attempted jugular approach initially. In the first case, we failed to pass the balloon catheter into the shunt because of significant strain on the renal vein preventing us from pushing the balloon catheter further (Figure 4). In the second case, a lateral shunt collateral from the splenic vein was embolized with coils through jugular approach; however, due to the strain on the renal vein, we were unable to pass balloon catheter into the shunt. In both these cases, femoral approach was successful.

Close

Figure 4

46-year-old female, known idiopathic cirrhosis, presented with active gastric variceal bleeding. (a) Contrast-enhanced computed tomography Coronal maximum intensity projection, ISD ∼2.7 cm. (b) Fluoroscopic image showing 5F Cobra catheter and Amplatz wire forming large arc from jugular access. Negotiation of 6F balloon catheter was difficult and strain noted on inferior vena cava and renal vein. (c) Venogram performed after inflating balloon. No collaterals were seen. Easy negotiation of shunt vein was done through the transfemoral approach with long straight sheath support and subsequent successful balloon-occluded retrograde transvenous obliteration.

Export to PPT

In two cases of recurrent hepatic encephalopathy with more than one shunt, femoral and jugular approaches were used for closer gastrorenal shunt and farther splenorenal shunt, respectively. One patient had two shunts, one gastrorenal and one splenorenal shunt. Gastrorenal shunt (ISD ∼3.13 cm) was accessed through femoral route and splenorenal shunt (ISD - 5.10 cm) was accessed through jugular route, and balloons were placed for embolization and successful BRTO was performed (Figure 5). Another patient had 3 shunts, one gastrorenal and two (proximal and distal) splenorenal shunts. Gastrorenal shunt (ISD - 3.38 cm) was accessed through femoral route and distal splenorenal shunt (ISD - 7.55 cm) was accessed through jugular route, and balloons were placed for embolization. However, proximal shunt (ISD - 5.09 cm) was accessed through femoral route after many unsuccessful attempts. We were unable to pass the balloon catheter due to wire instability, so we had to embolize the shunt with 4F vertebral catheter and 035’ coils (Figure 6).

Close

Figure 5

70-year-old male, known hepatitis B virus cirrhosis, presented with recurrent hepatic encephalopathy. (a) Contrast-enhanced computed tomography coronal maximum intensity projection image showing gastrorenal and splenorenal shunts with ISD ∼3.13 and 5.10 cm, respectively. Previously, TACE was performed for ruptured segment 6 HCC with complete response to treatment (arrow). (b) Fluoroscopic image showing balloon catheters in both shunts. (c) Fluroscopic image after completion of BRTO procedure showing contrast staining of shunts.

Export to PPT

Close

Figure 6

69-year-old male, known idiopathic liver cirrhosis, presented with recurrent hepatic encephalopathy. (a) Contrast-enhanced computed tomography oblique Coronal maximum intensity projection (MIP) image showing gastrorenal (open arrow) and splenorenal (arrow) shunts with ISD ∼3.32 and 5.09 cm, respectively. (b) CECT oblique coronal MIP image showing proximal (arrow) and distal (arrow heads) splenorenal shunts with distance between them measuring 2.46 cm. (c) Fluoroscopic image showing two microcatheters in the gastrorenal shunt by femoral approach and distal splenorenal shunt by jugular approach. Shunt venogram performed through distal splenorenal shunt (arrow heads). Coils were placed in collateral. (d) Fluoroscopic image showing contrast staining of distal splenorenal shunt after placing coils and balloon. Proximal splenorenal shunt venogram after accessing through femoral route (arrow). Difficult cannulation by transfemoral route. (e) Inability to pass balloon catheter (due to wire instability) into the proximal splenorenal shunt; therefore, it was embolized with coils (arrow). (f) Fluoroscopic image showing two balloons in gastrorenal (open arrow) and distal splenorenal shunts and sclerosant staining of shunts.

Export to PPT

In the remaining cases, ISD was between 2.67 cm and 3.5 cm. Balloon catheter negotiation across the renal vein into the shunt was easier with ISD <3 cm compared to a patient with ISD >3 cm (Figure 7). In cases with ISD >3 cm, the glide wire tends to slip back into IVC or renal vein (Figures 6 and 8). Once a stiff wire was negotiated into the shunt, it was easier to advance the balloon catheter into the shunt. After balloon inflation, large collaterals were identified and embolized with micro coils (4 cases of bleeds and 1 case of hepatic encephalopathy).

Close

Figure 7

63-year-old male, known alcoholic cirrhosis, presented with active gastric variceal bleeding. (a) The ISD measures 2.67 cm on the contrast-enhanced computed tomography coronal maximum intensity projection image. (b) Fluoroscopic image taken after inflation of the occlusion balloon with contrast medium, and the injection of contrast medium fills the large shunt. A phrenic collateral vein had been embolized with coils. Note that the sheath tip is placed in the left renal vein.

Export to PPT

Close

Figure 8

65-year-old male, known hepatitis B virus cirrhosis, presented with active gastric variceal bleeding. (a) Contrast-enhanced computed tomography coronal maximum intensity projection image with ISD ∼3.5 cm. (b) Fluoroscopic image showing abnormal curves of 5F catheter with glide wire in distal aspect of shunt and difficulty negotiating the shunt. (c) Fluoroscopic image after balloon inflation. Sheath tip in the shunt. This case demonstrates the concept that the larger the distance of the shunt from inferior vena cava, the more difficult it is to cannulate the shunt from the femoral route.

Export to PPT

Post-procedure, all the patients with refractory hepatic encephalopathy showed improvement. By day 3, the encephalopathy had improved, and by 1 month, reduction in serum ammonia levels was noted.

The procedure was technically successful in all the patients with no immediate procedure-related complications. With the exception of a small minor contrast leak, no major venous rupture occurred. Focal minor renal vein thrombosis occurred around the catheter in two patients, which resolved without any further intervention. All the patients were transferred to surgical ICU for post-procedure monitoring. Two patients died during follow-up: One due to liver failure 3 days after the procedure and the other due to sepsis, liver, and multiorgan failure at 3 months. All the remaining patients were subsequently discharged between post-procedure days 5 and 7.

DISCUSSION

Gastric varices are often difficult to manage and require multidisciplinary management. Gastric variceal bleeding tends to require increased blood transfusion requirements and have higher mortality rate compared to esophageal variceal bleed.^[4] BRTO has become an established procedure in the management of gastric varices. It has a technical success rate of around 91%.^[5]

Obtaining a stable balloon position is vital to prevent migration of the embolic agent into the renal vein. Transfemoral approach is preferred in view of “pushability” of the wire-catheter system and safety.^[6] Given the variation in the availability of equipment, operators use what equipment is locally. Hence, it is important to plan the procedure, particularly the access, as jugular approach can increase the procedure time by nearly 1 h.^[6]

Transfemoral catheterization of the gastrorenal shunt can be challenging. In cases when the left renal vein makes an acute angle with the IVC, the transjugular approach is more favorable.^[6,7] If the shunt is relatively perpendicular to the renal vein, a femoral approach is recommended. If the angle is relatively acute with the renal vein, a transjugular approach is recommended.^[6] Saad et al. stated that the further the distance of the shunt from the IVC, the greater the preference for the transjugular approach.^[6] However, they did not quantify the distance between the shunt and IVC. According to Gwon et al., transjugular approach was used when the angle was acute between left renal vein and adrenal vein.^[8]

When the authors used long curved sheaths, they considered the angle between IVC and renal vein and between the renal/adrenal vein and shunt to prevent failed catheterization or instability of the balloon. However, after using long, curved support sheaths, we feel ISD matters more than the angles as sheaths allow greater stability, regardless of angles. Graphical representation shows transfemoral and transjugular approach in passing wire and catheters with strain placed on the renal vein (Figure 9a-d).

Close

Figure 9

Effect of ISD on sheath/wire placement through transjugular (a and b) and transfemoral (c and d) approach. Longer the ISD (double-headed arrows), easier the cannulation of shunt through jugular approach (a) and difficult through femoral approach (c) and vice versa (b and d).

Export to PPT

An acute angle between the renal vein and shunt can make BRTO technically challenging. To overcome this problem, Shukla et al. modified the technique using two jugular sheaths. A 5-F vascular sheath was placed from the right internal jugular vein and used for injecting sclerosant into the shunt through a coaxially introduced microcatheter. A balloon occlusion catheter was placed into the shunt coaxially through another 6-F vascular sheath placed from the external jugular vein.^[9] We feel that this technique should be used when microcatheter is a must for targeted embolization and balloon catheters are of smaller diameter. If a curved supporting sheath is used, an acute angle between renal vein and shunt matters less.

There is insufficient evidence to support angles determining in deciding which access route should be selected for BRTO. If adequate straight/curved supporting sheath is used during the BRTO procedure, ISD matters more than angles. Future pooled data from multiple centers can improve our understanding. Further studies are also needed to see if the ISD of 3.5cm can be used as a cut of to decide the approach for undertaking BRTO.

CONCLUSION

Pre-planning CECT with coronal MIP reformation with the measurement of the distance between the IVC and shunt can help decide whether the femoral or jugular approach should be used for BRTO.

On the basis of the technique used in our study, the jugular approach should be used when the ISD is >3.5 cm and the femoral approach when the ISD is <3.0–3.5 cm. More prospective data will help investigate our theory.