Visualization of gastric varices using angiographic C-arm CT during retrograde transvenous sclerotherapy

Abstract During retrograde transvenous sclerotherapy for gastric varices, sufficient opacification of the target varices on venography is essential for successful treatment. However, venography sometimes cannot identify target varices due to overlapping adjacent collateral vessels or leakage of contrast medium to other outflow veins. We report how C-arm CT images acquired using a flat-panel detector angiography system helped to identify target varices and predict the distribution of a sclerosant, which resulted in safer sclerotherapy and increased operator confidence.


Introduction
Angiographic C-arm CT has recently been used for vascular intervention because it can acquire images of the vascular anatomy and provide information about the surrounding soft tissues. [1][2][3][4][5] The C-arm fl at-panel angiography system provides multiplanar soft -tissue images as well as conventional planar angiograms from a single unit. Here, we assessed whether C-arm CT during retrograde transvenous sclerotherapy of gastric varices added additional useful information with regard to identifi cation of gastric varices and prediction of the extent of sclerosant distribution in a single patient.

Case Report
A 64-year-old man with liver cirrhosis related to hepatitis C was referred to our hospital with gastric varices. Contrastenhanced multidetector CT (MDCT) showed gastric varices directly connected to a dilated left adrenal vein. Endoscopy revealed that the varices were located at the fornix of the stomach. First, we performed CT during arterial portography (CTAP) to assess the target varices and their infl ow and outfl ow veins, using a 16-MDCT scanner (Somatom Sensation, Siemens Medical Solutions, Forchheim, Germany) with the following scanning parameters: 120 kV; 182 mAs; beam collimation, 0.75 mm; helical pitch, 1.15 mm; and rotation table speed, 0.5 s. Images were reconstructed in 5 mm-thick transverse sections to provide contiguous sections. A 4-Fr angiography catheter was placed in the proximal superior mesenteric artery before CTAP and through this iodinated contrast medium (90 ml) was delivered by a power injector at a rate of 3 ml/s. Image acquisition began at 25 s aft er starting contrast injection. This procedure revealed that the varices were approximately 3 cm in diameter and located at the fornix of the stomach [ Figure 1a]. The major aff erent veins were the left gastric vein, the posterior gastric vein, and the short gastric vein; the left adrenal vein was the only major drainage vessel. Because no other drainage veins were identifi ed, retrograde transvenous sclerotherapy for the gastric varices was scheduled 35 days later; we planned to use a C-arm angiography system with a 30 ϫ 30 cm fl atpanel detector (Innova 3100, GE Healthcare, Waukesha, Wisconsin) set at the following parameters: Total scanning angle, 200°; rotation speed, 20°/s; matrix size, 1500 ϫ 1500; isotropic voxel size, 0.2 mm; and eff ective fi eld of view, 18 cm 2 . Raw data sets were transferred to an external workstation (Advantage Workstation 4.2, GE Healthcare), where images were reconstructed over a period of about 2.25 min to produce multisectional images. We positioned a 6.5-Fr balloon catheter (Artec Balloon Catheter, Create Medic, Yokohama, Japan) at the proximal left adrenal vein via the right femoral vein, and a left adrenal venogram was obtained by manually injecting contrast medium (300 mg of iodine; Iopamidol) while the Abstract During retrograde transvenous sclerotherapy for gastric varices, suffi cient opacifi cation of the target varices on venography is essential for successful treatment. However, venography sometimes cannot identify target varices due to overlapping adjacent collateral vessels or leakage of contrast medium to other outfl ow veins. We report how C-arm CT images acquired using a fl atpanel detector angiography system helped to identify target varices and predict the distribution of a sclerosant, which resulted in safer sclerotherapy and increased operator confi dence.   Figure 1b]. To confi rm that the probable varices were the actual targets, contrast-enhanced C-arm CT images were obtained by manually injecting the same volume of contrast medium (up to 100 mg of iodine) diluted with saline while the balloon was reinfl ated. The C-arm CT delineated the target gastric varices with the same confi guration as the CTAP [ Figure 1a and c]. Meanwhile, the probable varices visualized on conventional planar venogram were identifi ed as the targets by referring to the corresponding C-arm CT coronal images [ Figure 1b and d]. Because the varices were suffi ciently opacifi ed, 40 ml of 5% ethanolamine oleate in iodinated contrast medium was injected through the balloon catheter and the occlusion balloon was kept infl ated for 1 h.
When the sclerosant was completely injected, unenhanced C-arm CT images immediately demonstrated total fi lling of the target varices with the sclerosant [ Figure 2a] and partial fi lling of the gastrorenal shunt [ Figure 2b]. Of the aff erent gastric veins, sclerosant completely fi lled the posterior gastric vein [ Figure 2b], but was not distributed to the left gastric vein. As much as possible of the sclerosant was manually aspirated. The balloon catheter was then carefully defl ated and removed under fl uoroscopic guidance. Contrast-enhanced MDCT was done 7 days aft er sclerotherapy for therapeutic evaluation. The distribution of the sclerosant visualized on unenhanced C-arm CT images obtained immediately aft er injection was comparable with that of the thrombosed gastric varices observed on post-therapeutic MDCT [ Figure 2a and c]. The MDCT also confi rmed that the posterior gastric vein was completely thrombosed [ Figure 2b and d], whereas the left gastric vein remained patent.

Discussion
Gastric varices with spontaneous gastrosystemic shunts have been recently treated by retrograde sclerotherapy through the outfl ow veins of the varices. [6][7][8] The outcome of treatment depends on proper visualization and identifi cation of the intended target varices using balloon-occluded retrograde venography through the gastrosystemic shunt. Gastric varices can sometimes be partially overlapped by surrounding dilated collaterals or are not opacifi ed due to leakage of contrast medium into the systemic circulation via other outfl ow vessels during conventional venography. Therefore, venography with diff erent projections is needed to confi rm whether target varices are suffi ciently visible and opacifi ed before treatment. C-arm angiography using a fl at-panel detector is a novel imaging modality that can generate both conventional planar angiograms and lowcontrast soft tissue images in multiple planes. C-arm CT for abdominal interventions has recently been clinically applied, [1][2][3][4][5]9] but its utility in balloon-occluded retrograde transvenous sclerotherapy for gastric varices has not been described.
In our patient we did C-arm CT, which provided additional information regarding the target varices, and this increased confi dence among the angiography operators. Although the target varices of this patient were subsequently visualized by venography, the possibility that they might not be opacifi ed due to excessive outfl ow into the gastrosystemic shunt could not be excluded using only venography. On the other hand, C-arm CT was useful for assessing the exact location and extent of opacifi cation of target varices, because it constantly visualizes the intended target varices with surrounding soft tissue structures similar to MDCT. Furthermore, unenhanced C-arm CT obtained immediately aft er injecting the sclerosant accurately can predict the extent of sclerosant distribution. This technique allows additional sclerosant injection before the varices become thrombosed if the sclerosant distribution within the target varices is insuffi cient on posttherapeutic C-arm CT images. Moreover, post-therapeutic C-arm CT and MDCT were found to be comparable in their ability to evaluate thrombosis of varices and their related veins and therefore it may be possible to avoid additional contrast-enhanced MDCT to evaluate the outcome of sclerotherapy.
The C-arm CT imaging described herein has several limitations. Our C-arm angiography system was equipped with a fl atpanel detector of 30 cm 2 that produces an eff ective fi eld of view of 18 cm 2 , which is quite limited and results in truncated images of the gastric veins or the gastrosystemic shunts. C-arm CT imaging requires longer setup and reconstruction times than conventional venography to produce images. Furthermore, the low-contrast nature of C-arm CT imaging requires administration of additional contrast medium for proper visualization of target varices and surrounding structures.

Conclusion
C-arm CT provides substantially more information about the target varices (with regard to identifi cation and opacifi cation) than does conventional venography; Like MDCT, it can visualize the target varices with surrounding soft -tissue structures during transvenous sclerotherapy of gastric varices. C-arm CT is also useful for predicting sclerosant distribution within target varices and related veins during injection procedures, which allows additional sclerosant injection before the varices become thrombosed if the sclerosant is insuffi ciently distributed within the targets. C-arm CT allows operators to assess therapeutic outcomes with increased confi dence.