Gadoxetic acid disodium-enhanced magnetic resonance imaging for biliary and vascular evaluations in preoperative living liver donors: Comparison with gadobenate dimeglumine-enhanced MRI

Authors


Abstract

Purpose

To compare gadoxetic acid disodium (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) with gadobenate dimeglumine (Gd-BOPTA)-enhanced MRI in preoperative living liver donors for the evaluation of vascular and biliary variations.

Materials and Methods

Sixty-two living liver donors who underwent preoperative MRI were included in this study. Thirty-one patients underwent MRI with Gd-EOB-DTPA enhancement, and the other 31 underwent MRI with Gd-BOPTA enhancement. Two abdominal radiologists retrospectively reviewed dynamic T1-weighted and T1-weighted MR cholangiography images and ranked overall image qualities for the depiction of the hepatic artery, portal vein, hepatic vein, and bile duct on a 5-point scale and determined the presence and types of normal variations in each dynamic phase. Semiquantitative analysis for bile duct visualization was also conducted by calculating bile duct-to-liver contrast ratios.

Results

No statistical differences were found between the two contrast media in terms of hepatic artery or bile duct image quality by the two reviewers, or in terms of portal vein image quality by one reviewer (P > 0.05). Gd-BOPTA provided better image qualities than Gd-EOB-DTPA for the depiction of hepatic veins by both reviewers, and for the depiction of portal veins by one reviewer (P < 0.01). The two contrast media-enhanced images had similar bile duct-to-liver contrast ratios (P > 0.05). Regarding diagnostic accuracies with hepatic vascular/biliary branching types, no significant differences were observed between the two contrast media (P > 0.05).

Conclusion

Gd-EOB-DTPA could be as useful as Gd-BOPTA for the preoperative evaluation of living liver donors, and has the advantage of early hepatobiliary phase image acquisition. J. Magn. Reson. Imaging 2011;33:149–159. © 2010 Wiley-Liss, Inc.

A GROWING DEMAND for liver transplantation and a shortage of cadaveric livers has increased the prevalence of adult living-donor liver transplantation. However, partial hepatectomy of healthy donors is accompanied by the risks of various vascular and biliary complications (2%–32%) in donors. Transplantation-related complications have also been reported in recipients, including biliary (6%–35%) and vascular (overall incidence of 9%) complications (1–7). Therefore, detailed preoperative evaluations of hepatic vascular and biliary anatomies are mandatory.

Multidetector computed tomography (MDCT) is the primary imaging tool used for the preoperative evaluation of liver donors because it provides fine and accurate details of the hepatic arterial anatomy like digital subtraction angiography and reliable information on portal vein and hepatic vein variations (8–10). However, MDCT is limited in terms of determining the presence of fatty liver and evaluating bile duct variants, which must be performed prior to liver donor surgery. Even though biliary CT contrast agents that enable CT cholangiography are available in some countries, their availabilities are limited. The continuous infusion of biliary contrast agent and the mandatory delayed scan time (about 30 minutes) is another limitation of the clinical use of CT cholangiography (8). In addition, some reports have suggested that biliary contrast agents are associated with higher risks of contrast agent-related adverse reactions (11, 12).

Magnetic resonance imaging (MRI) has been used to complement CT during the evaluation of liver donors, especially for fatty liver and bile duct variant evaluations. Recently, due to technical advances in the MRI and contrast agent fields, MR angiography (MRA) has been reported to be as accurate as conventional angiography or even CT angiography at detecting surgically important anatomic variants of the hepatic vasculature in potential living liver donor candidates (13–16). The introductions of hepatobiliary agents, such as gadobenate dimeglumine (Gd-BOPTA) and mangafodipir trisodium (Mn-DPDP), have encouraged some researches to investigate if hepatobiliary agents can provide comprehensive assessments of the biliary and hepatic vascular systems and of the hepatic parenchyma in potential living liver donors (2, 17, 18).

Recently, gadoxetic acid disodium (Gd-EOB-DTPA), a new hepatobiliary agent, has become increasingly popular for liver imaging mainly because of its improved lesion-to-liver contrast caused by its greater hepatocyte specificity. This advantage manifests as a reduction in the time between injection and hepatobiliary phase image acquisition and markedly increases the rate of biliary excretion as compared with Gd-BOPTA (19–21). Early opacification of the biliary ducts can be a positive feature for the evaluation of bile duct anatomy in liver donors. However, early liver enhancement can reduce contrast resolution of the portal and hepatic veins. In addition, compared with Gd-BOPTA, the use of only 25% of the amount of gadolinium means that Gd-EOB-DTPA-enhanced MRI may show lower contrast resolution for the evaluation of the vascular anatomy. However, to the best of our knowledge, no study has evaluated the value of Gd-EOB-DTPA in preoperative living liver donors. Accordingly, the purpose of this study was to compare Gd-EOB-DTPA and Gd-BOPTA-enhanced MRI for the evaluation of vascular and biliary variations in preoperative living liver donors.

MATERIALS AND METHODS

The Institutional Review Board of our hospital approved this retrospective study and the requirement for informed consent was waived.

Patient Selection

From March 2007 to October 2007, 45 potential living liver donors underwent Gd-EOB-DTPA-enhanced (Primovist; Schering Pharma, Berlin, Germany) MRI. Of these, 14 were excluded because they did not later undergo liver surgery (Table 1). Finally, 31 patients were included in the Gd-EOB-DTPA-enhanced MRI group (mean age, 30.7 ± 9.5 years; M:F = 22:9).

Table 1. Causes of Not Undergoing Donor Hepatectomy in Potential Living Liver Donors
CausesNumber of CasesSpecific cause
  1. DDLT, dead donor liver transplantation; IHD, intrahepatic ducts; HLA, human leukocyte antigen.

Donors with gadoxetic acid-enhanced MRI
 Conversion to DDLT4 
 Severe fatty liver1Macrovesicular change 30%, microvesicular change 40%
 The preoperative death of recipients41 case: Penicilamine-related pancytopenia
1 case: Liver failure
2 cases: P. jiroveci pneumonia
 Donors' psychological problem31 case: Anxiety
2 cases: Depression
 Complex biliary anatomy on preoperative imaging1Right bile duct anatomic variation (+):
Right anterior segmental IHD originates from the common hepatic duct and then B7 IHD and left IHD bifurcates.
B6 IHD originates from left IHD.
 Others1Another donor (who was included our study) donated liver for an unknown cause.
Donors with gadobenate dimeglumine-enhanced MRI
 Conversion to DDLT6 
 Severe fatty liver1Macrovesicular change 30%, microvesicular change 40%
 The preoperative death of recipients1Terminal hepatic failure (hopeless discharge)
 Donors' psychological problem1Depression
 HLA mismatching1Antibody titer was not sufficiently decreased even after plasmapheresis.
 Transferred to another hospital for operation1Due to the want of the donor
 Others3Unknown personal problem

Another 45 potential living liver donors underwent Gd-BOPTA-enhanced (MultiHance; Bracco Imaging, Milan, Italy) MRI from October 2006 to February 2007 and from November 2007 to December 2007. Fourteen of these were also excluded for the same reason (Table 1). Finally, 31 were included in the Gd-BOPTA group (mean age ± SD, 31.4 ± 10.3 years; M:F = 23:8). Patients from this group were enrolled during two separate periods, as only Gd-EOB-DTPA was used in our hospital as a contrast medium for liver donor protocol MRI from March 2007 to October 2007. Ultimately, 62 patients that had undergone one of the two contrast-enhanced preoperative MRI studies were included in this study.

Table 2. Comparison of the Overall Image Quality of Each Phase
 PhaseGd-EOB-DTPAGD-BOPTAP-value
  1. Data are the average of scores of each patient image ± standard deviation.

Reviewer 1Hepatic arterial4.2 ± 0.93.8 ± 1.00.086
Portal venous3.8 ± 0.94.9 ± 0.3< 0.001
Hepatic venous2.3 ± 1.14.5 ± 1.0< 0.001
Hepatobiliary3.0 ± 0.93.4 ± 1.20.094
Reviewer 2Hepatic arterial4.8 ± 0.44.6 ± 0.50.383
Portal venous4.9 ± 0.24.9 ± 0.21.000
Hepatic venous3.7 ± 0.74.9 ± 0.4< 0.001
Hepatobiliary4.5 ± 0.84.2 ± 1.00.285

MR Examination

All images were obtained on a 1.5 T MR scanner (Magnetom Sonata; Siemens Medical Systems, Erlangen, Germany) using high-performance gradients (maximum amplitude, 40 mT/m; slew rate, 200 T/m/s) and eight receiver channels. A torso-phased array surface coil covering the entire liver was used in all examinations. Gd-BOPTA was administered at 0.1 mmol/kg (0.2 mL/kg body weight, i.v.), and Gd-EOB-DTPA was administered at 0.025 mmol/kg (0.1 mL/kg body weight, i.v.). Both contrast agents were administered at an injection rate of 2 mL/s using an MR-compatible power injector (Spectris, Medrad, Pittsburgh, PA) and these administrations were followed by a 20-mL saline chaser injected at 2 mL/s.

MR Sequences and Parameters

Donor protocol MRI consisted of baseline MR images, MR angiography (MRA), and MR cholangiopancreatography (MRCP). Baseline MRI included respiratory-triggered T2-weighted true fast imaging with steady-state precession (true-FISP) sequence, a T2-weighted half-Fourier acquisition single-shot turbo spin-echo (HASTE) sequence, and a breathold T1-weighted fast low-angle shot (FLASH) sequence with in-phase and opposed-phase imaging.

For MRA, a coronal 3D interpolated fat-suppressed FLASH sequence was obtained after administering either Gd-BOPTA or Gd-EOB-DTPA. The imaging parameters for this sequence were: TR/TE, 2.5/1.0; flip angle, 20°; matrix, 256 × 230; field of view, 380 × 380 mm; slice thickness, 1.5 mm; slice number, 64; an intermittent fat-saturation pulse; and parallel imaging (using the generalized autocalibrating partially parallel acquisition [GRAPPA] algorithm and an acceleration factor R of 2). The hepatic arterial phase was obtained at the time of left ventricle filling with contrast media (≈10–20 seconds after injection) under MR fluoroscopic guidance. During the 50 seconds following hepatic arterial phase imaging, multiple coronal sequential image sets were acquired every 10 seconds, so that a total of six sets of coronal images were obtained after contrast media injection for the evaluation of portal vein and hepatic vein anatomy (2). Additional axial and coronal T1 3D spoiled gradient echo fat-suppressed volume-interpolated breath-hold examination (VIBE) images were obtained at 120 seconds (axial) and 180 seconds (coronal), respectively, after injection for additional evaluations of hepatic vein anatomy. Baseline T2-weighted MR images (true-FISP) also showed hepatic veins with acceptable image quality, but these images were not included in the evaluation because the purpose of this study was to compare the diagnostic performances of the two contrast media.

For biliary anatomy depiction, delayed (hepatobiliary)-phase images (T1-weighted MR cholangiography, T1W-MRC) and axial and subsequent coronal images (obtained using the VIBE sequence) were acquired at 60 minutes after GD-BOPTA administration or at 20 minutes after Gd-EOB-DTPA administration (8, 9). Before GD-BOPTA delayed phase imaging, the patients were removed from the MR scanner because of the wait involved. The parameters used to acquire axial images were: TR/TE, 4.9/2.3; flip angle, 30°; slice thickness, 2.5 mm; slice number, 72; matrix, 256 × 175, and the parameters used to acquire coronal images were: TR/TE, 5.5/2.7; flip angle, 30°; slice thickness, 2.5 mm; slice number, 48; and matrix, 320 × 192.

Delayed (hepatobiliary)-phase axial and coronal images were also used for evaluating portal and hepatic vein anatomy because venous channels were well visualized on delayed-phase images with good negative contrast due to prominent hepatic parenchymal uptake of contrast media.

Heavily T2-weighted thick slab turbo spin-echo MRC (2D T2W-MRC) images and coronal respiratory-triggered fat-suppressed 3D T2-weighted turbo spin-echo MRC (3D T2W-MRC) images were also obtained before contrast media administration, as required by our institutional MRC protocol. However, only T1W-MRC images were used for image interpretation in the present study, as the purpose of our study was the comparison of diagnostic performance between the two contrast media.

Image Interpretation

Two experienced abdominal radiologists using a picture archiving and communication systems (PACS) monitor interpreted all MR images independently for overall image quality and the presence and type of vascular or biliary variations. These two radiologists were not involved in the clinical reading of liver donor MR images of the enrolled patients. At the time of interpretation, both radiologists were blinded to the surgical plan, surgical results, and preoperative CT findings. Sample images were ordered randomly, regardless of the type of contrast medium, and the reviewers were also blinded to any image information, such as examination dates and scan times, which could have provided clues to the type of contrast medium used.

The image qualities of the MR images obtained during each dynamic phase (hepatic artery, portal vein, hepatic vein, and hepatobiliary phases) were assessed using a 5-point scale (1, nondiagnostic; 2, poor; 3, fine; 4, good; 5, excellent). MRA image quality was evaluated by assessing the brightness and contrast of vessels, artifacts, and the visualization of surrounding anatomic structures. Image quality in the hepatobiliary phase was evaluated by assessing bile duct brightness, hepatic parenchymal enhancement, contrast between hepatic parenchyma and the bile duct, and artifact.

To evaluate hepatic artery anatomy, types of hepatic arterial branching patterns were categorized into six categories: conventional branching, a replaced right hepatic artery (RHA) from the superior mesenteric artery (SMA), an accessory RHA from the SMA, a replaced left hepatic artery (LHA) from the left gastric artery (LGA); an accessory LHA from the LGA; and an accessory middle hepatic artery (MHA) from the RHA. The reviewers determined which category of hepatic artery branching patterns was present. If two or more categories of variations existed simultaneously in one donor, the reviewers were requested to indicate all categories seen. In addition, if the reviewers encountered variations that could not be classified into one of the five above-described categories, they were requested to describe the variations. Accuracy defined as the proportion of cases in which anatomy had been correctly determined. Accuracy was calculated for each contrast medium-enhanced image and for each reviewer.

Evaluations of portal vein anatomy (22, 23) (Fig. 1) and biliary tree anatomy (23, 24) (Fig. 2) were conducted using a similar method, and possible variations were classified using previously published classification schemes (22–24). To classify anatomy of the hepatic vein, the number of accessory right inferior hepatic veins (RIHVs) greater than 5 mm in diameter was counted.

Figure 1.

Portal vein branching patterns. RAS = right anterior segmental portal vein supplying segments V and VIII; RPS = right posterior segmental portal vein supplying segments VI and VII (adapted from Arti M, et al. Intrahepatic portal venous variations: prevalence with US. Radiology 1992;184:157–158). a–f: Conventional branching (a), trifurcation, defined as a simultaneous origin of the RAS and RPS from the main portal vein (b), RPS arising from the main portal vein (c), RAS arising from the left portal vein (d), complete absence of the right portal vein (e), and absence of the horizontal segment of the left portal vein (f).

Figure 2.

Intrahepatic biliary duct branching patterns were divided into six types. Aberrant ducts are indicated in gray. C = cystic duct, Acc = accessory duct (adapted from Choi JW, et al. Anatomic variation in intrahepatic bile ducts: an analysis of intraoperative cholangiograms in 300 consecutive donors for living donor liver transplantation. Korean J Radiol 2003;4:85–90). a: The diagram shows Type 1, conventional branching. b: The diagram shows Type 2, which involves a triple confluence of the right anterior segmental duct that drains segments V and VIII (RAS), the right posterior segmental duct that drains segments VI and VII (RPS), and the left hepatic duct (LHD) into the common hepatic duct (CHD). c–e: The diagram shows Types 3A, 3B, and 3C: the RPS drains anomalously into the LHD, CHD, or cystic duct, respectively. f: The diagram shows Type 4: the right hepatic duct (RHD) drains into the cystic duct. g,h: The diagram shows Types 5A and 5B: the right accessory duct drains into the CHD or RHD. i: The diagram shows Type 6: segments II and III drain individually into the RHD or CHD.

To evaluate portal and hepatic vein anatomy, hepatobiliary phase images were provided to the reviewers with individual MRA phase images, because venous channels were well visualized on hepatobiliary phase images with good negative contrast due to prominent hepatic parenchymal contrast uptake. This evaluation method was given to provide an interpretational environment similar to that found in the clinical setting.

To determine whether hepatic venous or hepatobiliary phases provided the best means of evaluating hepatic vein anatomy in each MRI set, hepatic venous phase images and hepatobiliary phase images were separately provided to the reviewers, who then analyzed hepatic venous anatomy using the criteria as described above.

Semiquantitative Analysis of Bile Duct Visualization

Semiquantitative analysis of bile duct visualization was performed by calculating the bile duct to liver contrast ratio (Cduct-liver) at the right hepatic duct (RHD), the left hepatic duct (LHD), and the common hepatic duct (CHD). A radiologist, unaware of the type of contrast medium used, scan time, or patient information performed operator-defined region-of-interest (ROI) measurements of mean signal intensity (SI) of the RHD, LHD, CHD, and of adjacent hepatic parenchyma on axial T1W-MRC images using a PACS monitor and Digital Imaging and Communications in Medicine (DICOM) image viewing software (M-view, INFINITT, Seoul, Korea). To measure the SI values of hepatic ducts, circular ROIs were drawn to encompass as much of the hepatic ducts as possible. Mean ROI sizes and standard deviations were: 0.42 ± 0.20 cm2 at the RHD, 0.39 ± 0.04 cm2 at the LHD, and 1.07 ± 0.86 cm2 at the CHD. For hepatic parenchyma, ROIs (17.56 ± 2.07 cm2) were set in an area devoid of focal SI changes, large vessels, and prominent artifacts at the level where the SI of each segment of the hepatic ducts was measured. Cduct-liver values were calculated using (SIduct − SIliver) / SIliver.

Operative Findings

All 62 donors in our study population underwent right hemihepatectomy. Surgical records, which were used as the gold standard, were thoroughly compared with MRI findings. However, in terms of hepatic vein and portal vein anatomy, surgical records were at times too brief for a side-by-side comparison with MRI findings. Accordingly, thin section multidetector-row CT (MDCT) images with maximal intensity projection images (MIPs) were added to the operative record as reference standards of portal vein and hepatic vein anatomy for 23 of the 62 donors (13 in the Gd-BOPTA group and 10 in the Gd-EOB-DTPA group). MDCT with angiography provides a well-established alternative method to conventional angiography (25).

Statistical Analysis

Comparisons between the two contrast-medium-enhanced MR image sets with respect to overall image quality (rated using a 5-point scale) was performed using the Mann–Whitney U-test for each dynamic phase and each reviewer. Interobserver agreements regarding the presence of hepatic vascular/biliary variations were determined by calculating kappa values for each phase and contrast medium (poor, κ = 0; slight, κ 0.01–0.2; fair, κ 0.21–0.40; moderate, κ 0.41–0.60; good, κ 0.61–0.80; and excellent, κ 0.81–1.00 agreement). The accuracies of the two contrast-medium-enhanced MR image sets for the diagnosis of normal vascular/biliary anatomy were compared using the chi-squared test with Yate's correction, and using Fisher's exact test for all four phases and both reviewers. Student's t-test was used to determine the significance of differences between Cduct-liver values of the two contrast medium-enhanced groups. For all comparisons, P-values of 0.05 or less were considered to indicate statistically significant differences.

RESULTS

Overall Image Quality (Table 2)

No significant differences in image quality were found between the two contrast media in hepatic arterial phase images (P > 0.05), although both reviewers scored Gd-EOB-DTPA-enhanced images slightly higher.

However, for portal venous phase images, Gd-BOPTA-enhanced images were scored as having better image quality than Gd-EOB-DTPA-enhanced images by reviewer 1 (P < 0.05), although not by reviewer 2 (P = 1.000).

For hepatic venous phase images, both reviewers concluded that Gd-BOPTA-enhanced images provided superior image quality for the depiction of hepatic veins (P < .001).

For hepatobiliary phase images, both reviewers concluded that the two types of contrast medium-enhanced images provided similar image qualities.

Semiquantitative Study of Bile Duct Visualization (Table 3)

The two sets of contrast-medium-enhanced images showed similar bile duct-to-liver contrast ratios.

Table 3. Comparison of Bile Duct to Liver Contrast Ratio
ParameterLevelContrast agentMeanP-value
  1. Cduct-liver, bile duct to liver contrast ratio; CHD, common hepatic duct; LHD, left main hepatic duct; RHD, right main hepatic duct.

Cduct-liverCHDGd-EOB-DTPA1.2190.148
GD-BOPTA1.421
LHDGd-EOB-DTPA.6080.052
GD-BOPTA.784
RHDGd-EOB-DTPA.7510.936
GD-BOPTA.743

Determination of the Presence of Normal Variations (Table 4)

Kappa values of interobserver agreement for the presence of normal variation ranged from good to excellent, except for portal vein and bile duct variations. The kappa value for type of portal vein variation was low or not applicable, due to the extremely high proportion of donors with a normal anatomy. The kappa value for type of bile duct variation was also low, probably because of differences of the reviewers in the recognition of slight anatomical difference between normal type and variants types.

Table 4. Kappa Values of Interobserver Agreement for the Presence of Normal Variation
StructureGd-EOB-DTPAGD-BOPTA
  1. N/A, not applicable.

Hepatic artery0.800.60
Portal vein0.46N/A
Hepatic vein0.830.92
Bile duct0.530.38

Diagnostic Accuracies of Hepatic Vascular/Biliary Anatomy

Detection of Normal Variations of the Hepatic Artery (Fig. 3)

Hepatic arterial branching variants were common and were observed in 38 (61%) of the 62 patients. Thirteen patients had a replaced RHA from the SMA, and 12 patients had a replaced LHA from the LGA. Five patients had both a replaced RHA and a replaced LHA. One patient had an accessory RHA from the SMA, three patients had an accessory LHA from the LGA, and nine patients had an MHA from the RHA. Five patients had uncategorized miscellaneous variations, such as an MHA from the gastroduodenal artery (GDA) (n = 1) or common hepatic artery (CHA) (n = 2), a proper hepatic artery before the GDA (n = 1), or a cystic artery from the LHA (n = 1). No significant difference was found between the two contrast types by either reviewer in terms of predicting the type of hepatic artery branching patterns (Table 5).

Figure 3.

Coronal 3D fat-suppressed spoiled gradient echo images obtained in the arterial phase after injecting 0.025 mmol/kg Gd-EOB-DTPA into a 23-year-old male liver donor (a) and after injection of 0.1 mmol/kg Gd-BOPTA into a 34-year-old female liver donor (b). Both images depict variations of the left hepatic artery (arrow) from the left gastric artery well.

Table 5. Diagnostic Accuracy for Specific Types of Normal Variations
 StructureGd-EOB-DTPAGD-BOPTAP-value
  1. Data are the accuracy values of each phase presented as a percentage.

Reviewer 1Hepatic artery77.474.20.998
Portal vein87.196.80.349
Hepatic vein96.893.51.000
Bile duct90.371.00.109
Reviewer 2Hepatic artery77.467.70.568
Portal vein93.596.80.989
Hepatic vein93.593.51.000
Bile duct93.583.90.428

Detection of Normal Variations of the Portal Vein (Fig. 4)

A normal branching pattern of the portal vein (Type A) was identified in 57 (92%) of the 62 patients. The frequencies of the other variation types were: type B, 3% (n = 2); type C, 3% (n = 2); type D, 2% (n = 1); and types E and F, none. One case of type B and one case of type C were detected in the Gd-EOB-DTPA-enhanced donor group, and the other three variant cases were detected in the Gd-BOPTA-enhanced donor group. No significant difference was found between the two contrast media in terms of predicting the type of portal vein branching patterns (P > 0.05 for both reviewers) (Table 5).

Figure 4.

Coronal 3D fat-suppressed spoiled gradient echo images acquired during the portal venous phase. An image obtained using Gd-EOB-DTPA administered to a 22-year-old male liver donor (a). An image obtained using Gd-BOPTA administered to a 38-year-old male liver donor (b). The two contrast media-enhanced images depict a variation of the early branching right posterior portal vein (arrows, Type C of Fig. 1) from the main portal vein equally well.

Detection of Hepatic Vein Variations (Fig. 5)

Of the 62 donors, 28 (45%) had no significant accessory RIHV. Twenty (32%) had at least one significantly large (≥5 mm) accessory RIHV. No significant difference was found between the two contrast media in terms of predicting the number of accessory RIHVs (P = 0.05 for both reviewers) (Table 5). As for the diagnostic accuracies of the hepatic venous phase and the hepatobiliary phase for the detection of the accessory hepatic vein using Gd-EOB-DTPA, the hepatobiliary phase was found to be significantly better than the hepatic venous phase by both reviewers (Table 6). For Gd-BOPTA, the hepatic venous phase was found to have greater diagnostic accuracy than the hepatobiliary phase for both reviewers, but the difference was not statically significant (Table 6).

Figure 5.

Sixty-second delayed coronal 3D interpolated fat-suppressed FLASH sequence images and 120-second delayed T1 3D spoiled gradient echo fat-suppressed volume interpolated breathold examination (VIBE) images, and hepatobiliary phase coronal and axial VIBE sequence images. Gd-EOB-DTPA-enhanced MR images were obtained from a 46-year-old male liver donor (a,c). Gd-BOPTA-enhanced MR images were obtained from a 29-year-old male liver donor (b,d). Sixty- and 120-second delayed images enhanced with Gd-EOB-DTPA (a), the accessory right inferior hepatic vein (arrow) was not well visualized as compared with the Gd-BOPTA-enhanced image (b). Higher hepatic parenchymal signal intensity made the differentiation difficult, as shown by the Gd-EOB-DTPA-enhanced image. However, for hepatobiliary phase images obtained after Gd-EOB-DTPA (c) or Gd-BOPTA (d) administration, the hepatic vein was well demarcated on both contrast enhanced images with a negative contrast effect of liver parenchyma because washout occurred in the hepatic vein.

Table 6. Diagnostic Accuracy Comparison for Detection of Accessory Hepatic Vein Between Hepatic Venous Phase and Hepatobiliary Phase
  Hepatic venous phaseDelayed phaseP-value
  1. Data are the accuracy values of each phase presented as a percentage.

Reviewer 1Gd-EOB-DTPA71.0 (22/31)96.8 (30/31)0.016
GD-BOPTA93.5 (29/31)90.3 (28/31)0.997
P-Value0.0470.599 
Reviewer 2Gd-EOB-DTPA71.0 (22/31)93.5 (29/31)0.047
GD-BOPTA93.5 (29/31)87.1 (27/31)0.672
P-value0.0470.672 

Detection of Bile Duct Variations (Fig. 6)

A normal branching pattern of the biliary tree (Type A) was identified in 47 (76%) of the 62 donors. The frequencies of the other variation types were: Type 2, 8% (n = 5); Type 3A, 10% (n = 6); Type 3B, 6% (n = 4); and other types (3C-6), none. Three cases of Type 2, two cases of Type 3A, and one case of Type 3B were detected in the Gd-EOB-DTPA-enhanced donor group, and the other nine variant cases were detected in the Gd-BOPTA-enhanced group. No significant difference was found between the two types of contrast medium-enhanced images with respect to predicting type of biliary branching patterns (P >0.05 for both reviewers) (Table 5).

Figure 6.

Coronal and sagittal 3D fat-suppressed spoiled gradient echo images acquired at 20 minutes after Gd-EOB-DTPA administration during the delayed phase in a 23-year-old male liver donor (a) and acquired at 60 minutes after Gd-BOPTA administration during the delayed phase in a 34-year-old female liver donor (c) (T1-MRC) are shown. Respiratory-triggered fat-suppressed T2-weighted turbo spin-echo MR cholangiographic (3D T2-MRC) images with MIP reconstruction of the same patients (b, Gd-EOB-DTPA administration; d, Gd-BOPTA administration). T1-MRC images using both contrast media, depicting well the bile duct trifurcation variation (arrows), are comparable to that achieved using conventional T2-MRC.

Regarding types of biliary anatomy misinterpreted by the reviewers, in the Gd-EOB-DTPA-enhanced group reviewer 1 misinterpreted biliary variants Types 3A and 3B as Type 1 (normal branching pattern) in two and one cases, respectively (accuracy, 90.3%) and reviewer 2 misinterpreted Type 1 as Type 2 in two cases (accuracy, 93.5%). In the Gd-BOPTA-enhanced group, reviewer 1 misinterpreted biliary anatomy in nine cases (accuracy, 71.0%): Type 1 as Type 2 (n = 2); Type 2 as Type 1 (n = 2); Type 3A as Type 1 (n = 4); and Type 3B as Type 1 (n = 1). Reviewer 2 misinterpreted biliary anatomy in five cases (accuracy, 83.9%): Type 1 as Type 2 (n = 1); Type 2 as Type 1 (n = 2); Type 3A as Type 1 (n = 1); and Type 3B as Type 1 (n = 1).

DISCUSSION

In this study, Gd-EOB-DTPA-enhanced MRI with MRA and MRC showed comparable image quality in the hepatic arterial, portal, and hepatobiliary phases and similar bile duct to liver contrast ratio in the hepatobiliary phase as Gd-BOPTA-enhanced MRI, which has been described as the sole preoperative imaging modality for the evaluation of potential living liver donors (2). Furthermore, the reviewers' accuracies for detecting correct types of anatomy of the hepatic vascular and biliary systems showed no significant difference between the two contrast medium-enhanced MR image sets.

In terms of image quality in the arterial phase, our results show that the two contrast medium-enhanced images were not significantly different. The two contrast media have similar, efficient T1 relaxivities: 6.7 (6.3–7.1) mM.s−1 for Gd-BOPTA and 7.3 (6.9–7.7) mM.s−1 for Gd-EOB-DTPA in the blood at 1.5 T (26). It was expected that image quality in the arterial phase would be better in the Gd-BOPTA-enhanced study than in the Gd-EOB-DTPA-enhanced study due to the 4 times greater dose of Gd-BOPTA administered, as the T1 relaxivities of the two contrast media in the blood are similar. However, interestingly, we found that the image quality of the arterial phase and the diagnostic accuracy of the hepatic artery anatomy were similar for the two contrast medium-enhanced image sets (Fig. 3).

Our results also show that the two contrast medium-enhanced image sets showed no significant differences in image quality during the hepatobiliary phase and similar accuracies in terms of biliary tree anatomy. In our semiquantitative analysis, bile duct-to-liver contrast ratios were also similar for the two contrast medium-enhanced image sets. This is probably because the biliary excretion ratio of Gd-EOB-DTPA (57.0 ± 2.49%) is much higher than that of Gd-BOPTA (3–5%) (27), and compensated for the 4-fold dosage difference. Given the fact that Gd-BOPTA has a much longer waiting time for the acquisition of the hepatobiliary phase, these findings indicate that Gd-EOB-DTPA provides advantages during biliary anatomy evaluations of liver donors and at a smaller dosage.

In the hepatic venous phase, Gd-BOPTA provided better image quality than Gd-EOB-DTPA, as determined by both reviewers. This can be explained by the pharmacokinetic difference between the two drugs. Gd-EOB-DTPA uptake into hepatocytes occurs more rapidly and is taken up by a higher proportion of cells than Gd-BOPTA, and it also has a faster absorption rate. A rhesus monkey study, in which an ROI analysis of normal liver parenchyma was performed, showed that Gd-EOB-DTPA provides a greater initial increase in the parenchymal SI of liver (151 ± 21%, mean ± 1 SD) than Gd-BOPTA (114 ± 45%) at 1 minute after injecting the same dosage (0.1 mmol/kg) (28). This finding indicates that the parenchymal SI of liver is higher for Gd-EOB-DTPA-enhanced images than Gd-BOPTA-enhanced images obtained during the hepatic venous phase (60-second delayed images after contrast medium injection). Since the earlier, higher enhancement of liver parenchyma of Gd-EOB-DTPA has a negative effect on contrast between enhanced hepatic veins and liver parenchyma, hepatic veins seem to be more difficult to distinguish from the high signal liver parenchyma on Gd-EOB-DTPA-enhanced images. Likewise, in the portal venous phase, the pharmacokinetic difference described above may partly explain at least why one of the two reviewers scored the image quality of Gd-EOB-DTPA lower than that of Gd-BOPTA.

In terms of the diagnostic accuracies of hepatic vascular/biliary anatomy, no significant difference was observed between the contrast agents by either reviewer. Furthermore, the poor image quality of the hepatic venous phase in Gd-EOB-DTPA-enhanced MR images was compensated for by excellent hepatic vein anatomy depiction on hepatobiliary phase images during investigations of hepatic vein anatomy. These findings indicate that Gd-EOB-DTPA may be as useful as Gd-BOPTA for the diagnosis of normal variations of the vascular and biliary systems in living liver donors, despite the differences between these contrast agents in terms of image quality.

In a previous study on the preoperative MRI of living liver donors using Gd-BOPTA, an accuracy of 79% was obtained for the diagnosis of hepatic arterial anatomy, of 100% for the diagnosis of portal venous anatomy, and of 96% for that of hepatic venous anatomy (2). The diagnostic accuracy of Gd-BOPTA in our study concurs with this previous study (Table 5).

Our study has several limitations. First, it is limited by its retrospective design and its scale, and thus, further larger-scale prospective studies are required.

Second, there was some potential for selection bias, because surgical records were used as reference standards, and thus, only donors who underwent surgery were included. Accordingly, some donors with a complex vascular or bile duct variation may have been excluded. For this reason, differences between the two contrast media in terms of diagnostic accuracy may not reflect differences in the general population.

Third, in cases with inadequate surgical records regarding portal vein and hepatic vein anatomy, we used MDCT findings as a complementary reference standard. Although it is well known that MDCT angiography offers a good alternative to conventional angiography (10, 25), we cannot exclude the possibility that minor portal veins and hepatic vein variations were missed or misinterpreted.

Fourth, our study was not a paired study, and no direct comparison of the two contrast agents was possible in the same patients. Thus, unknown patient characteristics could have affected our image quality results and the confidences and accuracies of normal variation detection.

Fifth, Cduct-liver might be affected by slight changes in the locations of the ROI cursor, because of the diminutive size of the bile duct. However, we did attempt to locate the ROI cursor in the duct to reduce this source of error.

Finally, our donor MRI protocol includes 2D- and 3D-T2W-MRC, which also visualize ductal structures. However, we did not provide the reviewers with T2W-MRC, because the study objective was to compare the two contrast media. It is known that isotropic T2-weighted MRCP (3D T2W-MRC) has been vastly improved, and is superior to Gd-BOPTA-enhanced T1W-MRC for visualizing branching ducts (29). To verify the diagnostic efficacies of Gd-EOB-DTPA-enhanced T1W-MRC by comparing with T2W-MRC, further study is required.

Summarizing, Gd-EOB-DTPA was found to be as useful as Gd-BOPTA for the preoperative evaluation of living liver donors in terms of the detection and identification of vascular and biliary variations, which is essentially required for accurate and safe living donor liver transplantation. It should also be noted that the use of Gd-EOB-DTPA can reduce the administered dosage appreciably as compared with Gd-BOPTA. Furthermore, because the use of Gd-EOB-DTPA reduces the time delay required to obtain hepatobiliary phase images, its use can increase MR unit throughput, and it is also more convenient for patients.

Ancillary