Between January 2000 and December 2003, 165 consecutive patients underwent orthotopic liver transplantation at our institution. Comprehensive liver MRI and gadolinium-enhanced 3D MR angiography were performed preoperatively in 98 of these patients to assess vascular anatomy, detect focal liver lesions, and identify sequelae of portal hypertension such as varices and enlarged venous collaterals. Five of the 98 patients were excluded because they had undergone a prior liver transplantation. Two adult patients who had undergone hepatic lobectomy for hepatocellular carcinoma (HCC) were also excluded. All three pediatric patients were excluded because their hepatic vessels were very small and the liver transplantation technique was different from that used in adults. Furthermore, four patients who had undergone aortic jump graft were also excluded because the arterial reconstruction was different and the diameter of the native common hepatic artery did not affect the incidence of hepatic artery complications (described in Liver Transplantation). The study group consisted of the remaining 84 patients, all of whom underwent preoperative MR angiography and liver transplantation. The institutional review board at our hospital approved this retrospective study.
Fifty-five males and 29 females who ranged from 13 to 69 years old (mean age, 52.4 years) were in the study group. The causes of end-stage liver disease and the indications for liver transplantation are summarized in Table 1. Hepatitis C (n = 16) was the most common cause, followed by alcoholic liver disease (n = 12). Eleven patients had both. Eleven of 84 patients had HCC. Of the 11 patients with HCC, five underwent radiofrequency ablation either before (n = 3) or after (n = 2) MR angiography. The number of HCC tumor nodules ranged from one to four (mean, 1.5), and the size of the largest nodule ranged from 1.4 to 4.5 cm (mean, 2.6 cm). All three patients with HCC tumor nodules larger than 3 cm had undergone radiofrequency ablation before MR angiography. One patient had a transjugular intrahepatic portosystemic shunt stent placed before MR angiography. However, the stent did not compromise depiction of the hepatic arteries. None of the patients in the study group underwent hepatic artery chemoembolization or coronary vein coil embolization.
The medical records and operative reports were reviewed to determine the type of hepatic artery anastomosis used for the liver transplant and the incidence of posttransplantation hepatic artery stenosis and thrombosis. The postoperative follow-up period ranged from 40 days to 3.7 years (mean, 1.5 years).
Gadolinium-enhanced 3D MR angiography was performed on a 1.5-T scanner (CV/i LX, GE Healthcare) equipped with a four-element torso coil. All MR angiography studies were part of a comprehensive liver examination that included both MRI and MR angiography. The delay between contrast injection and image acquisition for the arterial phase of MR angiography was determined using a contrast-bolus timing run through an oblique sagittal segment of the mid abdominal aorta. Three milliliters of gadodiamide (Omniscan, Nycomed) followed by a 25 mL of saline flush was injected into the antecubital vein at 3 mL/sec. MR angiography was a 3D fast spoiled gradient-echo sequence performed with an axial (n = 75), coronal (n = 4), or sagittal (n = 5) volume and the following parameters: TR range/TE range, 4–5/0.8–1.3 (fractional echo); flip angle, 15–20°; bandwidth, 83 kHz; matrix, 256 x 128–160; partition thickness, 3–4 mm; partitions, 30–50; field of view, 34–44 x 25–33 cm; signal average, 1; zero-filling in the slice direction, 2 times; sequential phase-encode ordering; and acquisition time, 19–25 sec. The parameters were adjusted to include the entire liver and the hepatic vasculature and to provide the highest spatial resolution within an acceptable breath-hold period for the patient. An unenhanced MR angiography data set was acquired to familiarize the patient with the breath-holding instructions, confirm appropriate placement of the imaging volume, and provide a mask for subtraction images. Subsequently, one arterial and two venous phase MR angiography data sets were acquired, each during end-inspiration. Gadodiamide at 0.2 mmol/kg was power-injected (Spectris, Medrad) through a 20- or 22-gauge IV at 3 mL/sec, followed by a 25-mL saline flush at the same rate.
All patients underwent cadaveric liver transplantation. The interval between MR angiography and liver transplantation ranged from 3 to 561 days (mean, 144 days). The standard technique at our institution includes a branch patch arterial anastomosis, which is almost always formed at the takeoff of the gastroduodenal artery from the common hepatic artery  (Fig. 1A). However, four patients required an infrarenal aortic jump graft, which was formed from the donor's iliac artery  (Fig. 1B). Of these four patients, two had variant hepatic artery anatomy and a small-caliber common hepatic artery (one had a replaced right hepatic artery, and the other had both a replaced right hepatic artery and an accessory left hepatic artery). A third patient had a severe stenosis of the native celiac artery, and the fourth patient had a replaced right hepatic artery and a moderate celiac stenosis. At our institution, a branch patch anastomosis is preferred to an aortic jump graft because the former is simpler, and the aortic jump graft is reserved for possible future retransplantation. The decision concerning whether to use an aortic graft is based on an intraoperative assessment of the quality and flow of the vessels. Doppler sonography is not routinely used to measure hepatic artery flow. Muiesan et al.  reported that the incidence of hepatic artery thrombosis in aortic jump grafts was similar to that in standard hepatic artery anastomoses.
The standard surgical technique was also modified whenever there was variant hepatic artery anatomy in the donor liver, including a replaced right hepatic artery from a superior mesenteric artery (n = 6), an accessory left hepatic artery (n = 3), a replaced left hepatic artery (n = 2), a replaced right hepatic artery and an accessory left hepatic artery (n = 2), an accessory right hepatic artery (n = 1), and a replaced right hepatic artery from the celiac trunk (the so-called double hepatic artery, n = 1). For donor livers with a replaced right hepatic artery, the hepatic artery anastomosis was formed in two parts: a primary anastomosis between the donor celiac artery with an aortic patch (called a Carrel patch) and the recipient branch patch at the gastroduodenal artery takeoff and a secondary anastomosis between the replaced right hepatic artery and the proximal stump of the donor splenic artery. For donor livers with a replaced or accessory left hepatic artery, a similar strategy was used, but only the primary anastomosis (donor celiac artery with aortic patch) was formed .
The MR angiography data were transferred to an off-line UNIX-based computer workstation (Advantage Windows, GE Healthcare) for review and postprocessing. MR angiograms were reviewed retrospectively using multiplanar reformations and maximum intensity projections (MIP) (full volume and targeted) by two of the authors to quantify the degree to which distal hepatic artery branches could be visualized. For this purpose, the following qualitative scale was developed: poor, visualization of the proper hepatic artery; fair, visualization of the main right and left hepatic arteries; good, visualization of the segmental branches of the right and left hepatic arteries; and excellent, visualization of the subsegmental branches of the right and left hepatic arteries. Any MR angiogram receiving a score of poor was considered unsuitable for evaluating variant hepatic artery anatomy. The retrospective assessment of hepatic artery anatomy was made by consensus between the two reviewers. In all cases, the surgical record served as the standard of reference for the presence and type of variant hepatic artery anatomy.
Two radiologists measured the diameter of the distal common hepatic artery near the origin of the gastroduodenal artery. In each case, the reviewers used targeted MIP images generated by each reviewer on the computer workstation to identify the gastroduodenal artery takeoff. The measurements were obtained twice on the computer workstation by using electronic calipers and the magnification tool. The average diameters were calculated and recorded.
The frequency of hepatic artery complications after liver transplantation was compared for patients having classic and variant hepatic artery anatomy. In classic anatomy, the right and left hepatic arteries arise from a single proper hepatic artery, which in turn arises from a single common hepatic artery, which arises from the celiac artery. The data were analyzed using Fisher's exact test; a p value of less than 0.05 was considered significant. The odds ratio and 95% confidence intervals were calculated. A 95% confidence interval for the odds ratio that does not include 1.0 indicates a statistically significant difference at a p value of 0.05 . The mean diameter of the distal common hepatic artery was calculated for patients having classic or variant hepatic artery anatomy and for patients who did or did not have hepatic artery complications after liver transplantation. These data were analyzed using a Student's t test; a p value of less than 0.05 was considered to be statistically significant.